Spaces:
Sleeping
Sleeping
Upload folder using huggingface_hub
Browse files
modules/slot_layout_planner/__init__.py
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Planning utilities for chart-only, slot-guided infographic generation."""
|
| 2 |
+
|
modules/slot_layout_planner/chart_sanitizer.py
ADDED
|
@@ -0,0 +1,502 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import argparse
|
| 4 |
+
import json
|
| 5 |
+
import re
|
| 6 |
+
from dataclasses import asdict, dataclass
|
| 7 |
+
from pathlib import Path
|
| 8 |
+
from typing import Iterable
|
| 9 |
+
|
| 10 |
+
from lxml import etree
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
SVG_NS = "http://www.w3.org/2000/svg"
|
| 14 |
+
REMOVABLE_SLOT_ATTRS = {
|
| 15 |
+
"data-asset-slot",
|
| 16 |
+
"data-reserved-slot",
|
| 17 |
+
"data-slot-id",
|
| 18 |
+
"data-slot-kind",
|
| 19 |
+
"data-slot-policy",
|
| 20 |
+
"data-asset-source-policy",
|
| 21 |
+
}
|
| 22 |
+
PROTECTED_TITLE_TOKENS = {
|
| 23 |
+
"axis-title",
|
| 24 |
+
"x-axis-title",
|
| 25 |
+
"y-axis-title",
|
| 26 |
+
"legend-title",
|
| 27 |
+
"chart-title",
|
| 28 |
+
}
|
| 29 |
+
REMOVABLE_CLASS_FRAGMENTS = (
|
| 30 |
+
"reserved-asset-slot",
|
| 31 |
+
"reference-title",
|
| 32 |
+
"decorative-image-slot",
|
| 33 |
+
"hero-photo",
|
| 34 |
+
"hero-image",
|
| 35 |
+
"topic-art",
|
| 36 |
+
"header-art",
|
| 37 |
+
"header-illustration",
|
| 38 |
+
"header-image",
|
| 39 |
+
"title-art-slot",
|
| 40 |
+
"illustration-slot",
|
| 41 |
+
"image-slot",
|
| 42 |
+
)
|
| 43 |
+
EXPLICIT_PROTECTED_ATTRS = (
|
| 44 |
+
"data-chart-internal",
|
| 45 |
+
"data-chart-protected",
|
| 46 |
+
"data-polisher-protected",
|
| 47 |
+
"data-data-bearing",
|
| 48 |
+
)
|
| 49 |
+
EXPLICIT_EDITABLE_ATTRS = (
|
| 50 |
+
"data-polisher-editable",
|
| 51 |
+
"data-planned-element",
|
| 52 |
+
"data-planned-slot",
|
| 53 |
+
)
|
| 54 |
+
EDITABLE_SLOT_FRAGMENTS = (
|
| 55 |
+
"header",
|
| 56 |
+
"hero",
|
| 57 |
+
"illustration",
|
| 58 |
+
"context",
|
| 59 |
+
"summary",
|
| 60 |
+
"callout",
|
| 61 |
+
"footer",
|
| 62 |
+
"bottom",
|
| 63 |
+
"title",
|
| 64 |
+
"subtitle",
|
| 65 |
+
"ornament",
|
| 66 |
+
"background",
|
| 67 |
+
"decorative",
|
| 68 |
+
"topic art",
|
| 69 |
+
"topic image",
|
| 70 |
+
"topic visual",
|
| 71 |
+
"primary visual",
|
| 72 |
+
"right column",
|
| 73 |
+
"lower right",
|
| 74 |
+
"insight",
|
| 75 |
+
)
|
| 76 |
+
PROTECTED_SLOT_FRAGMENTS = (
|
| 77 |
+
"point marker",
|
| 78 |
+
"data point",
|
| 79 |
+
"bubble point",
|
| 80 |
+
"bubble icon",
|
| 81 |
+
"bubble identity",
|
| 82 |
+
"plain bubble",
|
| 83 |
+
"bar logo",
|
| 84 |
+
"bar icon",
|
| 85 |
+
"bar icon slot",
|
| 86 |
+
"bar category",
|
| 87 |
+
"bar cap",
|
| 88 |
+
"card header icon",
|
| 89 |
+
"row icon",
|
| 90 |
+
"row badge",
|
| 91 |
+
"row logo",
|
| 92 |
+
"row source",
|
| 93 |
+
"row country",
|
| 94 |
+
"row group",
|
| 95 |
+
"row topic",
|
| 96 |
+
"category icon",
|
| 97 |
+
"category badge",
|
| 98 |
+
"category logo",
|
| 99 |
+
"legend icon",
|
| 100 |
+
"legend marker",
|
| 101 |
+
"axis icon",
|
| 102 |
+
"series icon",
|
| 103 |
+
"group icon",
|
| 104 |
+
"metric icon",
|
| 105 |
+
"field icon",
|
| 106 |
+
"item icon",
|
| 107 |
+
"item icon slot",
|
| 108 |
+
"platform icon",
|
| 109 |
+
"temporal icon",
|
| 110 |
+
"year card",
|
| 111 |
+
"perimeter category",
|
| 112 |
+
"stacked area series",
|
| 113 |
+
"donut center",
|
| 114 |
+
"rose sector",
|
| 115 |
+
"rose center",
|
| 116 |
+
"technique icon",
|
| 117 |
+
"unit symbol",
|
| 118 |
+
"measure icon",
|
| 119 |
+
"measure badge",
|
| 120 |
+
"measure pill",
|
| 121 |
+
"badge trend mark",
|
| 122 |
+
"histogram trend mark",
|
| 123 |
+
"classification chip",
|
| 124 |
+
"material icon",
|
| 125 |
+
"rank badge",
|
| 126 |
+
"source logo",
|
| 127 |
+
"subtitle icon",
|
| 128 |
+
"trend icon",
|
| 129 |
+
"annotation icon",
|
| 130 |
+
"cue endpoint",
|
| 131 |
+
)
|
| 132 |
+
PROTECTED_CONTEXT_FRAGMENTS = (
|
| 133 |
+
"data point",
|
| 134 |
+
"point layer",
|
| 135 |
+
"point mark",
|
| 136 |
+
"bubble point",
|
| 137 |
+
"plot area",
|
| 138 |
+
"plot layer",
|
| 139 |
+
"axis",
|
| 140 |
+
"legend",
|
| 141 |
+
"series",
|
| 142 |
+
)
|
| 143 |
+
DATA_BEARING_ATTRS = {
|
| 144 |
+
"data-category",
|
| 145 |
+
"data-date",
|
| 146 |
+
"data-group",
|
| 147 |
+
"data-index",
|
| 148 |
+
"data-key",
|
| 149 |
+
"data-label",
|
| 150 |
+
"data-name",
|
| 151 |
+
"data-rank",
|
| 152 |
+
"data-raw-value",
|
| 153 |
+
"data-series",
|
| 154 |
+
"data-series-name",
|
| 155 |
+
"data-time",
|
| 156 |
+
"data-unit",
|
| 157 |
+
"data-value",
|
| 158 |
+
"data-values",
|
| 159 |
+
"data-x",
|
| 160 |
+
"data-y",
|
| 161 |
+
"data-y0",
|
| 162 |
+
"data-y1",
|
| 163 |
+
"data-y2",
|
| 164 |
+
"data-year",
|
| 165 |
+
}
|
| 166 |
+
DATA_BEARING_ROLE_VALUES = {
|
| 167 |
+
"data",
|
| 168 |
+
"data mark",
|
| 169 |
+
"data node",
|
| 170 |
+
"data point",
|
| 171 |
+
"value",
|
| 172 |
+
"value node",
|
| 173 |
+
"value-node",
|
| 174 |
+
}
|
| 175 |
+
|
| 176 |
+
|
| 177 |
+
@dataclass
|
| 178 |
+
class RemovedNode:
|
| 179 |
+
tag: str
|
| 180 |
+
reason: str
|
| 181 |
+
class_name: str
|
| 182 |
+
slot_name: str
|
| 183 |
+
|
| 184 |
+
|
| 185 |
+
@dataclass
|
| 186 |
+
class SanitizeReport:
|
| 187 |
+
input_svg: str
|
| 188 |
+
output_svg: str
|
| 189 |
+
removed_count: int
|
| 190 |
+
removed_by_reason: dict[str, int]
|
| 191 |
+
removed_nodes: list[RemovedNode]
|
| 192 |
+
|
| 193 |
+
|
| 194 |
+
def _local_name(elem: etree._Element) -> str:
|
| 195 |
+
return etree.QName(elem).localname if isinstance(elem.tag, str) else ""
|
| 196 |
+
|
| 197 |
+
|
| 198 |
+
def _class_text(elem: etree._Element) -> str:
|
| 199 |
+
return str(elem.get("class") or "")
|
| 200 |
+
|
| 201 |
+
|
| 202 |
+
def _class_tokens(elem: etree._Element) -> set[str]:
|
| 203 |
+
return {part.strip().lower() for part in re.split(r"\s+", _class_text(elem)) if part.strip()}
|
| 204 |
+
|
| 205 |
+
|
| 206 |
+
def _slot_name(elem: etree._Element) -> str:
|
| 207 |
+
for attr in ("data-asset-slot", "data-reserved-slot", "data-slot-id", "data-slot-kind"):
|
| 208 |
+
value = elem.get(attr)
|
| 209 |
+
if value:
|
| 210 |
+
return str(value)
|
| 211 |
+
return ""
|
| 212 |
+
|
| 213 |
+
|
| 214 |
+
def _has_slot_attr(elem: etree._Element) -> bool:
|
| 215 |
+
return any(elem.get(attr) is not None for attr in REMOVABLE_SLOT_ATTRS)
|
| 216 |
+
|
| 217 |
+
|
| 218 |
+
def _truthy_attr(elem: etree._Element, attrs: Iterable[str]) -> bool:
|
| 219 |
+
for attr in attrs:
|
| 220 |
+
value = elem.get(attr)
|
| 221 |
+
if value is None:
|
| 222 |
+
continue
|
| 223 |
+
if str(value).strip().lower() not in {"", "0", "false", "no"}:
|
| 224 |
+
return True
|
| 225 |
+
return False
|
| 226 |
+
|
| 227 |
+
|
| 228 |
+
def _normalized_text(value: str) -> str:
|
| 229 |
+
return re.sub(r"[^a-z0-9]+", " ", value.lower()).strip()
|
| 230 |
+
|
| 231 |
+
|
| 232 |
+
def _slot_context_text(elem: etree._Element) -> str:
|
| 233 |
+
parts = [
|
| 234 |
+
_slot_name(elem),
|
| 235 |
+
_class_text(elem),
|
| 236 |
+
str(elem.get("id") or ""),
|
| 237 |
+
str(elem.get("data-slot-anchor") or ""),
|
| 238 |
+
str(elem.get("data-asset-anchor") or ""),
|
| 239 |
+
str(elem.get("data-anchor") or ""),
|
| 240 |
+
str(elem.get("data-collision-rule") or ""),
|
| 241 |
+
str(elem.get("data-collision-policy") or ""),
|
| 242 |
+
str(elem.get("data-asset-collision") or ""),
|
| 243 |
+
str(elem.get("data-asset-source-policy") or ""),
|
| 244 |
+
str(elem.get("data-slot-policy") or ""),
|
| 245 |
+
]
|
| 246 |
+
return _normalized_text(" ".join(parts))
|
| 247 |
+
|
| 248 |
+
|
| 249 |
+
def _ancestor_context_text(elem: etree._Element) -> str:
|
| 250 |
+
parts = []
|
| 251 |
+
parent = elem.getparent()
|
| 252 |
+
while parent is not None:
|
| 253 |
+
parts.append(_class_text(parent))
|
| 254 |
+
parts.append(str(parent.get("id") or ""))
|
| 255 |
+
parent = parent.getparent()
|
| 256 |
+
return _normalized_text(" ".join(parts))
|
| 257 |
+
|
| 258 |
+
|
| 259 |
+
def is_chart_internal_slot(elem: etree._Element) -> bool:
|
| 260 |
+
"""Return True for asset slots that are part of chart semantics.
|
| 261 |
+
|
| 262 |
+
Many templates historically used ``data-asset-slot`` for two different
|
| 263 |
+
things: editable decorative placeholders and data-bearing icons inside the
|
| 264 |
+
chart. The polisher must not remove or mask the latter.
|
| 265 |
+
"""
|
| 266 |
+
|
| 267 |
+
if not _has_slot_attr(elem):
|
| 268 |
+
return False
|
| 269 |
+
if _truthy_attr(elem, EXPLICIT_PROTECTED_ATTRS):
|
| 270 |
+
return True
|
| 271 |
+
if _truthy_attr(elem, EXPLICIT_EDITABLE_ATTRS):
|
| 272 |
+
return False
|
| 273 |
+
|
| 274 |
+
text = _slot_context_text(elem)
|
| 275 |
+
if "input asset" in text or "preserve" in text or "original asset" in text:
|
| 276 |
+
return True
|
| 277 |
+
if any(fragment in text for fragment in PROTECTED_SLOT_FRAGMENTS):
|
| 278 |
+
return True
|
| 279 |
+
if any(fragment in text for fragment in EDITABLE_SLOT_FRAGMENTS):
|
| 280 |
+
return False
|
| 281 |
+
|
| 282 |
+
ancestor_text = _ancestor_context_text(elem)
|
| 283 |
+
if any(fragment in ancestor_text for fragment in PROTECTED_CONTEXT_FRAGMENTS):
|
| 284 |
+
return True
|
| 285 |
+
return False
|
| 286 |
+
|
| 287 |
+
|
| 288 |
+
def _has_data_bearing_attr(elem: etree._Element) -> bool:
|
| 289 |
+
for attr in DATA_BEARING_ATTRS:
|
| 290 |
+
value = elem.get(attr)
|
| 291 |
+
if value is not None and str(value).strip():
|
| 292 |
+
return True
|
| 293 |
+
role = elem.get("data-role")
|
| 294 |
+
if role and _normalized_text(str(role)) in DATA_BEARING_ROLE_VALUES:
|
| 295 |
+
return True
|
| 296 |
+
return False
|
| 297 |
+
|
| 298 |
+
|
| 299 |
+
def _subtree_has_data_bearing_attr(elem: etree._Element) -> bool:
|
| 300 |
+
if _has_data_bearing_attr(elem):
|
| 301 |
+
return True
|
| 302 |
+
return any(_has_data_bearing_attr(child) for child in elem.xpath(".//*"))
|
| 303 |
+
|
| 304 |
+
|
| 305 |
+
def _is_editable_asset_slot(elem: etree._Element) -> bool:
|
| 306 |
+
if not _has_slot_attr(elem):
|
| 307 |
+
return False
|
| 308 |
+
if _truthy_attr(elem, EXPLICIT_PROTECTED_ATTRS):
|
| 309 |
+
return False
|
| 310 |
+
if _truthy_attr(elem, EXPLICIT_EDITABLE_ATTRS):
|
| 311 |
+
return True
|
| 312 |
+
|
| 313 |
+
text = _slot_context_text(elem)
|
| 314 |
+
if "input asset" in text or "preserve" in text or "original asset" in text:
|
| 315 |
+
return False
|
| 316 |
+
if any(fragment in text for fragment in PROTECTED_SLOT_FRAGMENTS):
|
| 317 |
+
return False
|
| 318 |
+
return any(fragment in text for fragment in EDITABLE_SLOT_FRAGMENTS)
|
| 319 |
+
|
| 320 |
+
|
| 321 |
+
def is_inside_chart_internal_slot(elem: etree._Element) -> bool:
|
| 322 |
+
if is_chart_internal_slot(elem):
|
| 323 |
+
return True
|
| 324 |
+
parent = elem.getparent()
|
| 325 |
+
while parent is not None:
|
| 326 |
+
if is_chart_internal_slot(parent):
|
| 327 |
+
return True
|
| 328 |
+
parent = parent.getparent()
|
| 329 |
+
return False
|
| 330 |
+
|
| 331 |
+
|
| 332 |
+
def _is_axis_or_legend_title(elem: etree._Element) -> bool:
|
| 333 |
+
tokens = _class_tokens(elem)
|
| 334 |
+
class_text = " ".join(tokens)
|
| 335 |
+
return bool(tokens & PROTECTED_TITLE_TOKENS) or "axis" in class_text or "legend" in class_text
|
| 336 |
+
|
| 337 |
+
|
| 338 |
+
def _remove_reason(
|
| 339 |
+
elem: etree._Element,
|
| 340 |
+
*,
|
| 341 |
+
remove_images: bool,
|
| 342 |
+
remove_title_like: bool,
|
| 343 |
+
) -> str | None:
|
| 344 |
+
tag = _local_name(elem)
|
| 345 |
+
class_text = _class_text(elem).lower()
|
| 346 |
+
tokens = _class_tokens(elem)
|
| 347 |
+
|
| 348 |
+
if is_inside_chart_internal_slot(elem):
|
| 349 |
+
return None
|
| 350 |
+
|
| 351 |
+
if _subtree_has_data_bearing_attr(elem):
|
| 352 |
+
return None
|
| 353 |
+
|
| 354 |
+
if _has_slot_attr(elem):
|
| 355 |
+
if _is_editable_asset_slot(elem):
|
| 356 |
+
return "slot_metadata"
|
| 357 |
+
return None
|
| 358 |
+
|
| 359 |
+
if remove_images and tag == "image":
|
| 360 |
+
return "image_node"
|
| 361 |
+
|
| 362 |
+
for fragment in REMOVABLE_CLASS_FRAGMENTS:
|
| 363 |
+
if fragment in class_text:
|
| 364 |
+
return f"class:{fragment}"
|
| 365 |
+
|
| 366 |
+
if remove_title_like and "title" in class_text and not _is_axis_or_legend_title(elem):
|
| 367 |
+
return "class:title_like"
|
| 368 |
+
|
| 369 |
+
if remove_title_like and tag in {"g", "text"}:
|
| 370 |
+
if {"header", "subtitle", "kicker", "headline"} & tokens:
|
| 371 |
+
return "class:header_text_like"
|
| 372 |
+
|
| 373 |
+
return None
|
| 374 |
+
|
| 375 |
+
|
| 376 |
+
def _remove_empty_groups(root: etree._Element) -> int:
|
| 377 |
+
removed = 0
|
| 378 |
+
changed = True
|
| 379 |
+
while changed:
|
| 380 |
+
changed = False
|
| 381 |
+
for elem in list(root.xpath(".//*[local-name()='g']")):
|
| 382 |
+
if len(elem) != 0:
|
| 383 |
+
continue
|
| 384 |
+
if (elem.text or "").strip():
|
| 385 |
+
continue
|
| 386 |
+
if any(
|
| 387 |
+
elem.get(attr) is not None
|
| 388 |
+
for attr in ("id", "clip-path", "mask", "filter", "transform")
|
| 389 |
+
):
|
| 390 |
+
continue
|
| 391 |
+
parent = elem.getparent()
|
| 392 |
+
if parent is None:
|
| 393 |
+
continue
|
| 394 |
+
parent.remove(elem)
|
| 395 |
+
removed += 1
|
| 396 |
+
changed = True
|
| 397 |
+
return removed
|
| 398 |
+
|
| 399 |
+
|
| 400 |
+
def sanitize_chart_svg(
|
| 401 |
+
input_svg: Path,
|
| 402 |
+
output_svg: Path,
|
| 403 |
+
*,
|
| 404 |
+
remove_images: bool = False,
|
| 405 |
+
remove_title_like: bool = True,
|
| 406 |
+
remove_empty_groups: bool = True,
|
| 407 |
+
report_path: Path | None = None,
|
| 408 |
+
) -> SanitizeReport:
|
| 409 |
+
parser = etree.XMLParser(remove_blank_text=False, recover=True, huge_tree=True)
|
| 410 |
+
tree = etree.parse(str(input_svg), parser)
|
| 411 |
+
root = tree.getroot()
|
| 412 |
+
removed_nodes: list[RemovedNode] = []
|
| 413 |
+
removed_by_reason: dict[str, int] = {}
|
| 414 |
+
|
| 415 |
+
for elem in list(root.xpath(".//*")):
|
| 416 |
+
reason = _remove_reason(
|
| 417 |
+
elem,
|
| 418 |
+
remove_images=remove_images,
|
| 419 |
+
remove_title_like=remove_title_like,
|
| 420 |
+
)
|
| 421 |
+
if not reason:
|
| 422 |
+
continue
|
| 423 |
+
parent = elem.getparent()
|
| 424 |
+
if parent is None:
|
| 425 |
+
continue
|
| 426 |
+
node = RemovedNode(
|
| 427 |
+
tag=_local_name(elem),
|
| 428 |
+
reason=reason,
|
| 429 |
+
class_name=_class_text(elem),
|
| 430 |
+
slot_name=_slot_name(elem),
|
| 431 |
+
)
|
| 432 |
+
removed_nodes.append(node)
|
| 433 |
+
removed_by_reason[reason] = removed_by_reason.get(reason, 0) + 1
|
| 434 |
+
parent.remove(elem)
|
| 435 |
+
|
| 436 |
+
if remove_empty_groups:
|
| 437 |
+
empty_removed = _remove_empty_groups(root)
|
| 438 |
+
if empty_removed:
|
| 439 |
+
removed_by_reason["empty_group"] = empty_removed
|
| 440 |
+
|
| 441 |
+
root.set("data-chart-sanitized", "true")
|
| 442 |
+
root.set("data-chart-sanitizer-version", "1")
|
| 443 |
+
output_svg.parent.mkdir(parents=True, exist_ok=True)
|
| 444 |
+
tree.write(
|
| 445 |
+
str(output_svg),
|
| 446 |
+
encoding="utf-8",
|
| 447 |
+
xml_declaration=False,
|
| 448 |
+
pretty_print=False,
|
| 449 |
+
)
|
| 450 |
+
|
| 451 |
+
report = SanitizeReport(
|
| 452 |
+
input_svg=str(input_svg),
|
| 453 |
+
output_svg=str(output_svg),
|
| 454 |
+
removed_count=len(removed_nodes),
|
| 455 |
+
removed_by_reason=removed_by_reason,
|
| 456 |
+
removed_nodes=removed_nodes,
|
| 457 |
+
)
|
| 458 |
+
if report_path:
|
| 459 |
+
report_path.parent.mkdir(parents=True, exist_ok=True)
|
| 460 |
+
payload = asdict(report)
|
| 461 |
+
report_path.write_text(json.dumps(payload, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 462 |
+
return report
|
| 463 |
+
|
| 464 |
+
|
| 465 |
+
def parse_args(argv: Iterable[str] | None = None) -> argparse.Namespace:
|
| 466 |
+
parser = argparse.ArgumentParser(description="Remove non-chart template artifacts from a rendered SVG.")
|
| 467 |
+
parser.add_argument("--input-svg", type=Path, required=True)
|
| 468 |
+
parser.add_argument("--output-svg", type=Path, required=True)
|
| 469 |
+
parser.add_argument("--report", type=Path, default=None)
|
| 470 |
+
parser.add_argument(
|
| 471 |
+
"--remove-unmarked-images",
|
| 472 |
+
action="store_true",
|
| 473 |
+
help="Also remove plain <image> nodes that lack explicit slot/decorative metadata.",
|
| 474 |
+
)
|
| 475 |
+
parser.add_argument("--keep-title-like", action="store_true")
|
| 476 |
+
return parser.parse_args(argv)
|
| 477 |
+
|
| 478 |
+
|
| 479 |
+
def main(argv: Iterable[str] | None = None) -> int:
|
| 480 |
+
args = parse_args(argv)
|
| 481 |
+
report = sanitize_chart_svg(
|
| 482 |
+
input_svg=args.input_svg,
|
| 483 |
+
output_svg=args.output_svg,
|
| 484 |
+
remove_images=args.remove_unmarked_images,
|
| 485 |
+
remove_title_like=not args.keep_title_like,
|
| 486 |
+
report_path=args.report,
|
| 487 |
+
)
|
| 488 |
+
print(
|
| 489 |
+
json.dumps(
|
| 490 |
+
{
|
| 491 |
+
"output_svg": report.output_svg,
|
| 492 |
+
"removed_count": report.removed_count,
|
| 493 |
+
"removed_by_reason": report.removed_by_reason,
|
| 494 |
+
},
|
| 495 |
+
ensure_ascii=False,
|
| 496 |
+
)
|
| 497 |
+
)
|
| 498 |
+
return 0
|
| 499 |
+
|
| 500 |
+
|
| 501 |
+
if __name__ == "__main__":
|
| 502 |
+
raise SystemExit(main())
|
modules/slot_layout_planner/planner.py
ADDED
|
@@ -0,0 +1,1177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import argparse
|
| 4 |
+
import base64
|
| 5 |
+
import html
|
| 6 |
+
import importlib.util
|
| 7 |
+
import json
|
| 8 |
+
import os
|
| 9 |
+
import re
|
| 10 |
+
from dataclasses import asdict, dataclass
|
| 11 |
+
from pathlib import Path
|
| 12 |
+
from typing import Any, Iterable
|
| 13 |
+
|
| 14 |
+
from PIL import Image, ImageDraw
|
| 15 |
+
|
| 16 |
+
from modules.asset_regenerator.asset_regenerator import _absolute_bbox, _element_transform, _load_svg
|
| 17 |
+
from modules.asset_regenerator.asset_regenerator import render_svg_to_png
|
| 18 |
+
from modules.slot_layout_planner.chart_sanitizer import (
|
| 19 |
+
is_inside_chart_internal_slot,
|
| 20 |
+
sanitize_chart_svg,
|
| 21 |
+
)
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
@dataclass
|
| 25 |
+
class SlotPlanItem:
|
| 26 |
+
id: str
|
| 27 |
+
kind: str
|
| 28 |
+
bbox_px: tuple[int, int, int, int]
|
| 29 |
+
may_overlap_chart: bool
|
| 30 |
+
element: str
|
| 31 |
+
anchor: str
|
| 32 |
+
collision_rule: str
|
| 33 |
+
asset_source_policy: str
|
| 34 |
+
slot_requirements: list[str]
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
@dataclass
|
| 38 |
+
class PlannedSlotPackage:
|
| 39 |
+
slot_plan: Path
|
| 40 |
+
planned_svg: Path
|
| 41 |
+
planned_png: Path
|
| 42 |
+
reference_map: Path
|
| 43 |
+
sanitized_svg: Path
|
| 44 |
+
sanitized_png: Path
|
| 45 |
+
sanitizer_report: Path
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
@dataclass
|
| 49 |
+
class ProtectedRegion:
|
| 50 |
+
id: str
|
| 51 |
+
reason: str
|
| 52 |
+
bbox_px: tuple[int, int, int, int]
|
| 53 |
+
source_slot: str
|
| 54 |
+
anchor: str = ""
|
| 55 |
+
collision_rule: str = ""
|
| 56 |
+
semantic: str = ""
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
GPT_LAYOUT_INSTRUCTIONS = """You are designing editable generation slots for a data infographic.
|
| 60 |
+
|
| 61 |
+
Return valid JSON only. Do not include markdown fences.
|
| 62 |
+
|
| 63 |
+
The chart image is already rendered and will be placed at chart_bbox_px on the final canvas.
|
| 64 |
+
Design 3 to 6 editable slots for GPT image editing around or near that chart. The slots should
|
| 65 |
+
make the final piece feel intentionally designed, not like a fixed template.
|
| 66 |
+
|
| 67 |
+
Hard constraints:
|
| 68 |
+
- Use the provided canvas coordinate system in pixels.
|
| 69 |
+
- Return only editable slots; do not return chart_internal_slots. They are collected separately.
|
| 70 |
+
- Avoid every chart_internal_slot bbox completely.
|
| 71 |
+
- Do not cover axis labels, numeric values, category labels, legends, or data marks unless the slot is
|
| 72 |
+
a small callout and may_overlap_chart is true.
|
| 73 |
+
- If allow_chart_overlap is false, every returned slot must avoid the chart_bbox_px.
|
| 74 |
+
- Keep text slots large enough for legible typography and image slots large enough for clear imagery.
|
| 75 |
+
- Do not invent exact numeric values in slot text.
|
| 76 |
+
|
| 77 |
+
Required JSON shape:
|
| 78 |
+
{
|
| 79 |
+
"background": "#F7F8F4",
|
| 80 |
+
"design_rationale": "short explanation",
|
| 81 |
+
"slots": [
|
| 82 |
+
{
|
| 83 |
+
"id": "short_snake_case_id",
|
| 84 |
+
"kind": "title|subtitle|image|callout|context|annotation|badge",
|
| 85 |
+
"bbox_px": [x, y, width, height],
|
| 86 |
+
"may_overlap_chart": false,
|
| 87 |
+
"element": "what the image model should generate inside this slot",
|
| 88 |
+
"anchor": "where it sits in the composition",
|
| 89 |
+
"collision_rule": "explicit preservation rule",
|
| 90 |
+
"asset_source_policy": "llm_generated_text|image2_generated|llm_generated_text_and_shape|llm_generated_text_and_image",
|
| 91 |
+
"slot_requirements": ["short requirement", "short requirement"]
|
| 92 |
+
}
|
| 93 |
+
]
|
| 94 |
+
}
|
| 95 |
+
"""
|
| 96 |
+
|
| 97 |
+
ALLOWED_SLOT_POLICIES = {
|
| 98 |
+
"llm_generated_text",
|
| 99 |
+
"image2_generated",
|
| 100 |
+
"llm_generated_text_and_shape",
|
| 101 |
+
"llm_generated_text_and_image",
|
| 102 |
+
}
|
| 103 |
+
|
| 104 |
+
|
| 105 |
+
def _read_json(path: Path | None) -> dict[str, Any]:
|
| 106 |
+
if not path or not path.is_file():
|
| 107 |
+
return {}
|
| 108 |
+
try:
|
| 109 |
+
return json.loads(path.read_text(encoding="utf-8"))
|
| 110 |
+
except Exception:
|
| 111 |
+
return {}
|
| 112 |
+
|
| 113 |
+
|
| 114 |
+
def _load_openai_config(
|
| 115 |
+
*,
|
| 116 |
+
api_key: str | None = None,
|
| 117 |
+
api_key_env: str = "OPENAI_API_KEY",
|
| 118 |
+
base_url: str | None = None,
|
| 119 |
+
) -> tuple[str | None, str | None]:
|
| 120 |
+
resolved_key = api_key or os.environ.get(api_key_env)
|
| 121 |
+
resolved_base_url = base_url or os.environ.get("OPENAI_BASE_URL")
|
| 122 |
+
config_path = Path(__file__).resolve().parents[2] / "config.py"
|
| 123 |
+
if config_path.exists():
|
| 124 |
+
try:
|
| 125 |
+
spec = importlib.util.spec_from_file_location("_slot_layout_planner_config", config_path)
|
| 126 |
+
if spec is not None and spec.loader is not None:
|
| 127 |
+
module = importlib.util.module_from_spec(spec)
|
| 128 |
+
spec.loader.exec_module(module)
|
| 129 |
+
if not resolved_key:
|
| 130 |
+
for name in ("openai_api_key", "api_key", "client_key"):
|
| 131 |
+
value = getattr(module, name, None)
|
| 132 |
+
if isinstance(value, str) and value.strip():
|
| 133 |
+
resolved_key = value.strip()
|
| 134 |
+
break
|
| 135 |
+
if not resolved_base_url:
|
| 136 |
+
for name in ("openai_base_url", "base_url"):
|
| 137 |
+
value = getattr(module, name, None)
|
| 138 |
+
if isinstance(value, str) and value.strip():
|
| 139 |
+
resolved_base_url = value.strip()
|
| 140 |
+
break
|
| 141 |
+
except Exception:
|
| 142 |
+
pass
|
| 143 |
+
return resolved_key, resolved_base_url
|
| 144 |
+
|
| 145 |
+
|
| 146 |
+
def _summarize_data_for_layout(data: dict[str, Any]) -> dict[str, Any]:
|
| 147 |
+
if not data:
|
| 148 |
+
return {}
|
| 149 |
+
summary: dict[str, Any] = {}
|
| 150 |
+
for key in ("titles", "metadata", "description", "insights", "unit", "source"):
|
| 151 |
+
value = data.get(key)
|
| 152 |
+
if value:
|
| 153 |
+
summary[key] = value
|
| 154 |
+
|
| 155 |
+
rows = data.get("data") or data.get("rows") or data.get("values")
|
| 156 |
+
if isinstance(rows, list):
|
| 157 |
+
summary["row_count"] = len(rows)
|
| 158 |
+
summary["sample_rows"] = rows[:8]
|
| 159 |
+
elif isinstance(rows, dict):
|
| 160 |
+
summary["data_keys"] = list(rows.keys())[:20]
|
| 161 |
+
|
| 162 |
+
for key, value in data.items():
|
| 163 |
+
if key in summary or key in {"data", "rows", "values"}:
|
| 164 |
+
continue
|
| 165 |
+
if len(summary) >= 10:
|
| 166 |
+
break
|
| 167 |
+
if isinstance(value, (str, int, float, bool)) or value is None:
|
| 168 |
+
summary[key] = value
|
| 169 |
+
elif isinstance(value, list):
|
| 170 |
+
summary[key] = value[:5]
|
| 171 |
+
elif isinstance(value, dict):
|
| 172 |
+
summary[key] = {k: value[k] for k in list(value.keys())[:8]}
|
| 173 |
+
return summary
|
| 174 |
+
|
| 175 |
+
|
| 176 |
+
def _title_hint(data: dict[str, Any]) -> str:
|
| 177 |
+
titles = data.get("titles") if isinstance(data.get("titles"), dict) else {}
|
| 178 |
+
metadata = data.get("metadata") if isinstance(data.get("metadata"), dict) else {}
|
| 179 |
+
for value in (
|
| 180 |
+
titles.get("main_title"),
|
| 181 |
+
titles.get("title"),
|
| 182 |
+
metadata.get("title"),
|
| 183 |
+
metadata.get("name"),
|
| 184 |
+
):
|
| 185 |
+
if value:
|
| 186 |
+
return str(value)
|
| 187 |
+
return "Generate a concise data-aware headline"
|
| 188 |
+
|
| 189 |
+
|
| 190 |
+
def _extract_response_text(response: Any) -> str:
|
| 191 |
+
output_text = getattr(response, "output_text", None)
|
| 192 |
+
if isinstance(output_text, str) and output_text.strip():
|
| 193 |
+
return output_text.strip()
|
| 194 |
+
parts: list[str] = []
|
| 195 |
+
for item in getattr(response, "output", []) or []:
|
| 196 |
+
for content in getattr(item, "content", []) or []:
|
| 197 |
+
text = getattr(content, "text", None)
|
| 198 |
+
if isinstance(text, str):
|
| 199 |
+
parts.append(text)
|
| 200 |
+
elif isinstance(text, dict):
|
| 201 |
+
value = text.get("value")
|
| 202 |
+
if isinstance(value, str):
|
| 203 |
+
parts.append(value)
|
| 204 |
+
text = "\n".join(parts).strip()
|
| 205 |
+
if not text:
|
| 206 |
+
raise RuntimeError("OpenAI layout response did not include text output.")
|
| 207 |
+
return text
|
| 208 |
+
|
| 209 |
+
|
| 210 |
+
def _parse_json_object(text: str) -> dict[str, Any]:
|
| 211 |
+
cleaned = text.strip()
|
| 212 |
+
if cleaned.startswith("```"):
|
| 213 |
+
cleaned = cleaned.strip("`")
|
| 214 |
+
if cleaned.lower().startswith("json"):
|
| 215 |
+
cleaned = cleaned[4:].strip()
|
| 216 |
+
try:
|
| 217 |
+
data = json.loads(cleaned)
|
| 218 |
+
except json.JSONDecodeError:
|
| 219 |
+
start = cleaned.find("{")
|
| 220 |
+
end = cleaned.rfind("}")
|
| 221 |
+
if start < 0 or end <= start:
|
| 222 |
+
raise
|
| 223 |
+
data = json.loads(cleaned[start : end + 1])
|
| 224 |
+
if not isinstance(data, dict):
|
| 225 |
+
raise RuntimeError("OpenAI layout response must be a JSON object.")
|
| 226 |
+
return data
|
| 227 |
+
|
| 228 |
+
|
| 229 |
+
def _clamp_slot(
|
| 230 |
+
x: float,
|
| 231 |
+
y: float,
|
| 232 |
+
w: float,
|
| 233 |
+
h: float,
|
| 234 |
+
canvas_w: int,
|
| 235 |
+
canvas_h: int,
|
| 236 |
+
) -> tuple[int, int, int, int]:
|
| 237 |
+
x0 = max(0, min(int(round(x)), canvas_w - 1))
|
| 238 |
+
y0 = max(0, min(int(round(y)), canvas_h - 1))
|
| 239 |
+
ww = max(8, min(int(round(w)), canvas_w - x0))
|
| 240 |
+
hh = max(8, min(int(round(h)), canvas_h - y0))
|
| 241 |
+
return x0, y0, ww, hh
|
| 242 |
+
|
| 243 |
+
|
| 244 |
+
def _intersection_area(a: tuple[int, int, int, int], b: tuple[int, int, int, int]) -> int:
|
| 245 |
+
ax, ay, aw, ah = a
|
| 246 |
+
bx, by, bw, bh = b
|
| 247 |
+
ix0 = max(ax, bx)
|
| 248 |
+
iy0 = max(ay, by)
|
| 249 |
+
ix1 = min(ax + aw, bx + bw)
|
| 250 |
+
iy1 = min(ay + ah, by + bh)
|
| 251 |
+
return max(0, ix1 - ix0) * max(0, iy1 - iy0)
|
| 252 |
+
|
| 253 |
+
|
| 254 |
+
def _choose_overlay_slot(
|
| 255 |
+
chart_bbox: tuple[int, int, int, int],
|
| 256 |
+
canvas_w: int,
|
| 257 |
+
canvas_h: int,
|
| 258 |
+
protected_regions: list[ProtectedRegion],
|
| 259 |
+
) -> tuple[int, int, int, int]:
|
| 260 |
+
chart_x, chart_y, chart_w, chart_h = chart_bbox
|
| 261 |
+
candidate_w = chart_w * 0.34
|
| 262 |
+
candidate_h = min(190, chart_h * 0.22)
|
| 263 |
+
candidates = [
|
| 264 |
+
(chart_x + chart_w * 0.56, chart_y + chart_h * 0.16),
|
| 265 |
+
(chart_x + chart_w * 0.08, chart_y + chart_h * 0.16),
|
| 266 |
+
(chart_x + chart_w * 0.56, chart_y + chart_h * 0.54),
|
| 267 |
+
(chart_x + chart_w * 0.08, chart_y + chart_h * 0.54),
|
| 268 |
+
(chart_x + chart_w * 0.33, chart_y + chart_h * 0.34),
|
| 269 |
+
]
|
| 270 |
+
slots = [
|
| 271 |
+
_clamp_slot(x, y, candidate_w, candidate_h, canvas_w, canvas_h)
|
| 272 |
+
for x, y in candidates
|
| 273 |
+
]
|
| 274 |
+
if not protected_regions:
|
| 275 |
+
return slots[0]
|
| 276 |
+
return min(
|
| 277 |
+
slots,
|
| 278 |
+
key=lambda bbox: (
|
| 279 |
+
sum(_intersection_area(bbox, region.bbox_px) for region in protected_regions),
|
| 280 |
+
abs((bbox[0] + bbox[2] / 2) - (chart_x + chart_w * 0.73)),
|
| 281 |
+
),
|
| 282 |
+
)
|
| 283 |
+
|
| 284 |
+
|
| 285 |
+
def _build_slots(
|
| 286 |
+
canvas_w: int,
|
| 287 |
+
canvas_h: int,
|
| 288 |
+
chart_bbox: tuple[int, int, int, int],
|
| 289 |
+
data: dict[str, Any],
|
| 290 |
+
allow_chart_overlap: bool,
|
| 291 |
+
protected_regions: list[ProtectedRegion] | None = None,
|
| 292 |
+
) -> list[SlotPlanItem]:
|
| 293 |
+
margin = max(56, int(canvas_w * 0.047))
|
| 294 |
+
chart_x, chart_y, chart_w, chart_h = chart_bbox
|
| 295 |
+
title_hint = _title_hint(data)
|
| 296 |
+
protected_regions = protected_regions or []
|
| 297 |
+
|
| 298 |
+
def slot(
|
| 299 |
+
slot_id: str,
|
| 300 |
+
kind: str,
|
| 301 |
+
bbox: tuple[int, int, int, int],
|
| 302 |
+
element: str,
|
| 303 |
+
anchor: str,
|
| 304 |
+
overlap: bool = False,
|
| 305 |
+
policy: str = "image2_generated",
|
| 306 |
+
requirements: list[str] | None = None,
|
| 307 |
+
) -> SlotPlanItem:
|
| 308 |
+
return SlotPlanItem(
|
| 309 |
+
id=slot_id,
|
| 310 |
+
kind=kind,
|
| 311 |
+
bbox_px=bbox,
|
| 312 |
+
may_overlap_chart=bool(overlap and allow_chart_overlap),
|
| 313 |
+
element=element,
|
| 314 |
+
anchor=anchor,
|
| 315 |
+
collision_rule=(
|
| 316 |
+
"may overlap and visually integrate with chart content"
|
| 317 |
+
if overlap and allow_chart_overlap
|
| 318 |
+
else "stay visually coherent with the chart without requiring pixel preservation"
|
| 319 |
+
),
|
| 320 |
+
asset_source_policy=policy,
|
| 321 |
+
slot_requirements=requirements or [],
|
| 322 |
+
)
|
| 323 |
+
|
| 324 |
+
overlay = _choose_overlay_slot(chart_bbox, canvas_w, canvas_h, protected_regions)
|
| 325 |
+
visual = _clamp_slot(canvas_w - margin - 360, margin + 26, 360, 260, canvas_w, canvas_h)
|
| 326 |
+
title = _clamp_slot(margin, margin, canvas_w - 2 * margin - 250, 210, canvas_w, canvas_h)
|
| 327 |
+
subtitle = _clamp_slot(margin, margin + 232, canvas_w * 0.60, 100, canvas_w, canvas_h)
|
| 328 |
+
bottom_y = min(canvas_h - margin - 180, chart_y + chart_h + 42)
|
| 329 |
+
bottom = _clamp_slot(margin, bottom_y, canvas_w - 2 * margin, 170, canvas_w, canvas_h)
|
| 330 |
+
|
| 331 |
+
return [
|
| 332 |
+
slot(
|
| 333 |
+
"hero_title",
|
| 334 |
+
"title",
|
| 335 |
+
title,
|
| 336 |
+
f"Generate a polished editorial headline. Title hint: {title_hint}",
|
| 337 |
+
"top-left headline area",
|
| 338 |
+
policy="llm_generated_text",
|
| 339 |
+
requirements=[
|
| 340 |
+
"text may be rewritten by the image model",
|
| 341 |
+
"keep the headline legible and data-aware",
|
| 342 |
+
],
|
| 343 |
+
),
|
| 344 |
+
slot(
|
| 345 |
+
"subtitle_context",
|
| 346 |
+
"subtitle",
|
| 347 |
+
subtitle,
|
| 348 |
+
"Generate a short explanatory subtitle or deck based on the chart data.",
|
| 349 |
+
"under the main headline",
|
| 350 |
+
policy="llm_generated_text",
|
| 351 |
+
requirements=["one or two concise lines", "avoid invented numeric values"],
|
| 352 |
+
),
|
| 353 |
+
slot(
|
| 354 |
+
"topic_visual",
|
| 355 |
+
"image",
|
| 356 |
+
visual,
|
| 357 |
+
"Generate a thematic editorial image or icon cluster that matches the data topic.",
|
| 358 |
+
"upper-right visual support area",
|
| 359 |
+
policy="image2_generated",
|
| 360 |
+
requirements=["no logos", "no watermarks", "avoid small unreadable text"],
|
| 361 |
+
),
|
| 362 |
+
slot(
|
| 363 |
+
"chart_overlay_callout",
|
| 364 |
+
"callout",
|
| 365 |
+
overlay,
|
| 366 |
+
"Generate a concise visual callout overlay highlighting the most important visible trend.",
|
| 367 |
+
"inside or near the chart plot area",
|
| 368 |
+
overlap=True,
|
| 369 |
+
policy="llm_generated_text_and_shape",
|
| 370 |
+
requirements=[
|
| 371 |
+
"may overlap chart if useful",
|
| 372 |
+
"do not fabricate exact values unless they are visible in the chart",
|
| 373 |
+
],
|
| 374 |
+
),
|
| 375 |
+
slot(
|
| 376 |
+
"bottom_context_band",
|
| 377 |
+
"context",
|
| 378 |
+
bottom,
|
| 379 |
+
"Generate supporting context, iconography, or a short narrative footer for the infographic.",
|
| 380 |
+
"below the chart",
|
| 381 |
+
policy="llm_generated_text_and_image",
|
| 382 |
+
requirements=["keep text brief", "do not add source logos or watermarks"],
|
| 383 |
+
),
|
| 384 |
+
]
|
| 385 |
+
|
| 386 |
+
|
| 387 |
+
def _safe_slot_id(raw_id: Any, kind: str, index: int) -> str:
|
| 388 |
+
text = str(raw_id or "").strip().lower()
|
| 389 |
+
text = re.sub(r"[^a-z0-9]+", "_", text).strip("_")
|
| 390 |
+
if not text:
|
| 391 |
+
text = f"{kind}_{index:02d}"
|
| 392 |
+
if not re.match(r"^[a-z]", text):
|
| 393 |
+
text = f"slot_{text}"
|
| 394 |
+
return text[:64]
|
| 395 |
+
|
| 396 |
+
|
| 397 |
+
def _slot_kind(raw_kind: Any) -> str:
|
| 398 |
+
kind = re.sub(r"[^a-z0-9_]+", "_", str(raw_kind or "context").strip().lower()).strip("_")
|
| 399 |
+
return kind or "context"
|
| 400 |
+
|
| 401 |
+
|
| 402 |
+
def _coerce_requirements(value: Any) -> list[str]:
|
| 403 |
+
if not isinstance(value, list):
|
| 404 |
+
return []
|
| 405 |
+
requirements = []
|
| 406 |
+
for item in value[:8]:
|
| 407 |
+
text = str(item).strip()
|
| 408 |
+
if text:
|
| 409 |
+
requirements.append(text[:220])
|
| 410 |
+
return requirements
|
| 411 |
+
|
| 412 |
+
|
| 413 |
+
def _slot_overlap_area(
|
| 414 |
+
bbox: tuple[int, int, int, int],
|
| 415 |
+
regions: list[ProtectedRegion],
|
| 416 |
+
) -> int:
|
| 417 |
+
return sum(_intersection_area(bbox, region.bbox_px) for region in regions)
|
| 418 |
+
|
| 419 |
+
|
| 420 |
+
def _nudge_slot_clear_of_regions(
|
| 421 |
+
bbox: tuple[int, int, int, int],
|
| 422 |
+
regions: list[ProtectedRegion],
|
| 423 |
+
canvas_w: int,
|
| 424 |
+
canvas_h: int,
|
| 425 |
+
) -> tuple[int, int, int, int] | None:
|
| 426 |
+
if not regions or _slot_overlap_area(bbox, regions) == 0:
|
| 427 |
+
return bbox
|
| 428 |
+
|
| 429 |
+
x, y, w, h = bbox
|
| 430 |
+
candidates = [bbox]
|
| 431 |
+
for region in regions:
|
| 432 |
+
rx, ry, rw, rh = region.bbox_px
|
| 433 |
+
candidates.extend(
|
| 434 |
+
[
|
| 435 |
+
_clamp_slot(x, ry - h - 12, w, h, canvas_w, canvas_h),
|
| 436 |
+
_clamp_slot(x, ry + rh + 12, w, h, canvas_w, canvas_h),
|
| 437 |
+
_clamp_slot(rx - w - 12, y, w, h, canvas_w, canvas_h),
|
| 438 |
+
_clamp_slot(rx + rw + 12, y, w, h, canvas_w, canvas_h),
|
| 439 |
+
]
|
| 440 |
+
)
|
| 441 |
+
|
| 442 |
+
clear_candidates = [candidate for candidate in candidates if _slot_overlap_area(candidate, regions) == 0]
|
| 443 |
+
if clear_candidates:
|
| 444 |
+
return min(
|
| 445 |
+
clear_candidates,
|
| 446 |
+
key=lambda candidate: abs(candidate[0] - x) + abs(candidate[1] - y),
|
| 447 |
+
)
|
| 448 |
+
return None
|
| 449 |
+
|
| 450 |
+
|
| 451 |
+
def _slots_from_layout_payload(
|
| 452 |
+
data: dict[str, Any],
|
| 453 |
+
*,
|
| 454 |
+
canvas_w: int,
|
| 455 |
+
canvas_h: int,
|
| 456 |
+
chart_bbox: tuple[int, int, int, int],
|
| 457 |
+
allow_chart_overlap: bool,
|
| 458 |
+
chart_internal_slots: list[ProtectedRegion],
|
| 459 |
+
) -> tuple[list[SlotPlanItem], str, dict[str, Any]]:
|
| 460 |
+
raw_slots = data.get("slots")
|
| 461 |
+
if not isinstance(raw_slots, list):
|
| 462 |
+
raise RuntimeError("GPT layout response is missing a slots array.")
|
| 463 |
+
|
| 464 |
+
slots: list[SlotPlanItem] = []
|
| 465 |
+
seen_ids: set[str] = set()
|
| 466 |
+
validation_notes: list[dict[str, Any]] = []
|
| 467 |
+
for index, raw in enumerate(raw_slots, 1):
|
| 468 |
+
if not isinstance(raw, dict):
|
| 469 |
+
validation_notes.append({"index": index, "status": "skipped", "reason": "slot is not an object"})
|
| 470 |
+
continue
|
| 471 |
+
|
| 472 |
+
raw_bbox = raw.get("bbox_px")
|
| 473 |
+
if not isinstance(raw_bbox, (list, tuple)) or len(raw_bbox) != 4:
|
| 474 |
+
validation_notes.append({"index": index, "status": "skipped", "reason": "invalid bbox_px"})
|
| 475 |
+
continue
|
| 476 |
+
try:
|
| 477 |
+
bbox = _clamp_slot(
|
| 478 |
+
float(raw_bbox[0]),
|
| 479 |
+
float(raw_bbox[1]),
|
| 480 |
+
float(raw_bbox[2]),
|
| 481 |
+
float(raw_bbox[3]),
|
| 482 |
+
canvas_w,
|
| 483 |
+
canvas_h,
|
| 484 |
+
)
|
| 485 |
+
except (TypeError, ValueError):
|
| 486 |
+
validation_notes.append({"index": index, "status": "skipped", "reason": "non-numeric bbox_px"})
|
| 487 |
+
continue
|
| 488 |
+
|
| 489 |
+
if bbox[2] < 40 or bbox[3] < 32:
|
| 490 |
+
validation_notes.append({"index": index, "status": "skipped", "reason": "bbox too small"})
|
| 491 |
+
continue
|
| 492 |
+
|
| 493 |
+
nudged_bbox = _nudge_slot_clear_of_regions(bbox, chart_internal_slots, canvas_w, canvas_h)
|
| 494 |
+
if nudged_bbox is None:
|
| 495 |
+
validation_notes.append(
|
| 496 |
+
{
|
| 497 |
+
"index": index,
|
| 498 |
+
"status": "skipped",
|
| 499 |
+
"reason": "overlaps chart_internal_slots",
|
| 500 |
+
"bbox_px": list(bbox),
|
| 501 |
+
}
|
| 502 |
+
)
|
| 503 |
+
continue
|
| 504 |
+
if nudged_bbox != bbox:
|
| 505 |
+
validation_notes.append(
|
| 506 |
+
{
|
| 507 |
+
"index": index,
|
| 508 |
+
"status": "adjusted",
|
| 509 |
+
"reason": "nudged clear of chart_internal_slots",
|
| 510 |
+
"from_bbox_px": list(bbox),
|
| 511 |
+
"to_bbox_px": list(nudged_bbox),
|
| 512 |
+
}
|
| 513 |
+
)
|
| 514 |
+
bbox = nudged_bbox
|
| 515 |
+
|
| 516 |
+
requested_overlap = bool(raw.get("may_overlap_chart"))
|
| 517 |
+
may_overlap_chart = requested_overlap and allow_chart_overlap
|
| 518 |
+
if not may_overlap_chart and _intersection_area(bbox, chart_bbox) > 0:
|
| 519 |
+
validation_notes.append(
|
| 520 |
+
{
|
| 521 |
+
"index": index,
|
| 522 |
+
"status": "skipped",
|
| 523 |
+
"reason": "overlaps chart while may_overlap_chart is false",
|
| 524 |
+
"bbox_px": list(bbox),
|
| 525 |
+
}
|
| 526 |
+
)
|
| 527 |
+
continue
|
| 528 |
+
|
| 529 |
+
kind = _slot_kind(raw.get("kind"))
|
| 530 |
+
slot_id = _safe_slot_id(raw.get("id"), kind, index)
|
| 531 |
+
if slot_id in seen_ids:
|
| 532 |
+
slot_id = f"{slot_id}_{index:02d}"
|
| 533 |
+
seen_ids.add(slot_id)
|
| 534 |
+
|
| 535 |
+
policy = str(raw.get("asset_source_policy") or "").strip()
|
| 536 |
+
if policy not in ALLOWED_SLOT_POLICIES:
|
| 537 |
+
policy = "llm_generated_text_and_image" if kind in {"context", "callout", "annotation"} else "image2_generated"
|
| 538 |
+
|
| 539 |
+
collision_rule = str(raw.get("collision_rule") or "").strip()
|
| 540 |
+
if not collision_rule:
|
| 541 |
+
collision_rule = (
|
| 542 |
+
"may overlap chart but must preserve all chart labels, marks, icons, and numeric values"
|
| 543 |
+
if may_overlap_chart
|
| 544 |
+
else "avoid chart content and chart-internal icon slots"
|
| 545 |
+
)
|
| 546 |
+
|
| 547 |
+
slots.append(
|
| 548 |
+
SlotPlanItem(
|
| 549 |
+
id=slot_id,
|
| 550 |
+
kind=kind,
|
| 551 |
+
bbox_px=bbox,
|
| 552 |
+
may_overlap_chart=may_overlap_chart,
|
| 553 |
+
element=str(raw.get("element") or f"Generate {kind} content for this infographic.").strip(),
|
| 554 |
+
anchor=str(raw.get("anchor") or "").strip(),
|
| 555 |
+
collision_rule=collision_rule,
|
| 556 |
+
asset_source_policy=policy,
|
| 557 |
+
slot_requirements=_coerce_requirements(raw.get("slot_requirements")),
|
| 558 |
+
)
|
| 559 |
+
)
|
| 560 |
+
|
| 561 |
+
if not slots:
|
| 562 |
+
raise RuntimeError("GPT layout response did not produce any valid editable slots.")
|
| 563 |
+
|
| 564 |
+
background = str(data.get("background") or "#F7F8F4").strip()
|
| 565 |
+
if not re.match(r"^#[0-9a-fA-F]{6}$", background):
|
| 566 |
+
background = "#F7F8F4"
|
| 567 |
+
meta = {
|
| 568 |
+
"design_rationale": str(data.get("design_rationale") or "").strip(),
|
| 569 |
+
"validation_notes": validation_notes,
|
| 570 |
+
}
|
| 571 |
+
return slots, background, meta
|
| 572 |
+
|
| 573 |
+
|
| 574 |
+
def _build_gpt_slots(
|
| 575 |
+
*,
|
| 576 |
+
output_dir: Path,
|
| 577 |
+
chart_png: Path,
|
| 578 |
+
canvas_w: int,
|
| 579 |
+
canvas_h: int,
|
| 580 |
+
chart_bbox: tuple[int, int, int, int],
|
| 581 |
+
data: dict[str, Any],
|
| 582 |
+
allow_chart_overlap: bool,
|
| 583 |
+
chart_internal_slots: list[ProtectedRegion],
|
| 584 |
+
model: str,
|
| 585 |
+
api_key: str | None,
|
| 586 |
+
api_key_env: str,
|
| 587 |
+
base_url: str | None,
|
| 588 |
+
timeout_seconds: float,
|
| 589 |
+
max_retries: int,
|
| 590 |
+
max_output_tokens: int,
|
| 591 |
+
) -> tuple[list[SlotPlanItem], str, dict[str, Any]]:
|
| 592 |
+
request_path = output_dir / "layout_request.json"
|
| 593 |
+
response_path = output_dir / "layout_response.json"
|
| 594 |
+
payload = {
|
| 595 |
+
"canvas": {"width": canvas_w, "height": canvas_h},
|
| 596 |
+
"chart_bbox_px": list(chart_bbox),
|
| 597 |
+
"allow_chart_overlap": allow_chart_overlap,
|
| 598 |
+
"title_hint": _title_hint(data),
|
| 599 |
+
"data_summary": _summarize_data_for_layout(data),
|
| 600 |
+
"chart_internal_slots": [
|
| 601 |
+
{
|
| 602 |
+
"id": region.id,
|
| 603 |
+
"source_slot": region.source_slot,
|
| 604 |
+
"bbox_px": list(region.bbox_px),
|
| 605 |
+
"anchor": region.anchor,
|
| 606 |
+
"collision_rule": region.collision_rule,
|
| 607 |
+
}
|
| 608 |
+
for region in chart_internal_slots
|
| 609 |
+
],
|
| 610 |
+
"required_output_shape": {
|
| 611 |
+
"background": "#RRGGBB",
|
| 612 |
+
"design_rationale": "short string",
|
| 613 |
+
"slots": [
|
| 614 |
+
{
|
| 615 |
+
"id": "short_snake_case_id",
|
| 616 |
+
"kind": "title|subtitle|image|callout|context|annotation|badge",
|
| 617 |
+
"bbox_px": [0, 0, 100, 100],
|
| 618 |
+
"may_overlap_chart": False,
|
| 619 |
+
"element": "generation instruction",
|
| 620 |
+
"anchor": "layout anchor",
|
| 621 |
+
"collision_rule": "preservation rule",
|
| 622 |
+
"asset_source_policy": "llm_generated_text|image2_generated|llm_generated_text_and_shape|llm_generated_text_and_image",
|
| 623 |
+
"slot_requirements": ["requirement"],
|
| 624 |
+
}
|
| 625 |
+
],
|
| 626 |
+
},
|
| 627 |
+
}
|
| 628 |
+
request_log = {
|
| 629 |
+
"agent": "slot_layout_planner",
|
| 630 |
+
"model": model,
|
| 631 |
+
"instructions": GPT_LAYOUT_INSTRUCTIONS,
|
| 632 |
+
"payload": payload,
|
| 633 |
+
"images": [{"label": "sanitized_chart", "path": str(chart_png)}],
|
| 634 |
+
}
|
| 635 |
+
request_path.write_text(json.dumps(request_log, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 636 |
+
|
| 637 |
+
resolved_key, resolved_base_url = _load_openai_config(
|
| 638 |
+
api_key=api_key,
|
| 639 |
+
api_key_env=api_key_env,
|
| 640 |
+
base_url=base_url,
|
| 641 |
+
)
|
| 642 |
+
if not resolved_key:
|
| 643 |
+
raise RuntimeError(
|
| 644 |
+
f"{api_key_env} is not set and config.py does not define api_key/client_key; GPT layout cannot run."
|
| 645 |
+
)
|
| 646 |
+
try:
|
| 647 |
+
from openai import OpenAI
|
| 648 |
+
except ImportError as exc:
|
| 649 |
+
raise RuntimeError("The openai package is required for GPT layout planning.") from exc
|
| 650 |
+
|
| 651 |
+
client_kwargs: dict[str, Any] = {
|
| 652 |
+
"api_key": resolved_key,
|
| 653 |
+
"timeout": timeout_seconds,
|
| 654 |
+
"max_retries": max_retries,
|
| 655 |
+
}
|
| 656 |
+
if resolved_base_url:
|
| 657 |
+
client_kwargs["base_url"] = resolved_base_url
|
| 658 |
+
client = OpenAI(**client_kwargs)
|
| 659 |
+
|
| 660 |
+
content: list[dict[str, Any]] = [
|
| 661 |
+
{
|
| 662 |
+
"type": "input_text",
|
| 663 |
+
"text": (
|
| 664 |
+
"Return valid JSON only. No markdown fences.\n\n"
|
| 665 |
+
f"Payload:\n{json.dumps(payload, ensure_ascii=False, indent=2)}"
|
| 666 |
+
),
|
| 667 |
+
},
|
| 668 |
+
{"type": "input_text", "text": "Image: sanitized_chart"},
|
| 669 |
+
{"type": "input_image", "image_url": _image_data_uri(chart_png)},
|
| 670 |
+
]
|
| 671 |
+
kwargs: dict[str, Any] = {
|
| 672 |
+
"model": model,
|
| 673 |
+
"instructions": GPT_LAYOUT_INSTRUCTIONS,
|
| 674 |
+
"input": [{"role": "user", "content": content}],
|
| 675 |
+
"max_output_tokens": max_output_tokens,
|
| 676 |
+
"text": {"format": {"type": "json_object"}},
|
| 677 |
+
}
|
| 678 |
+
response = client.responses.create(**kwargs)
|
| 679 |
+
response_meta = {
|
| 680 |
+
"model": getattr(response, "model", model),
|
| 681 |
+
"response_id": getattr(response, "id", None),
|
| 682 |
+
}
|
| 683 |
+
text = _extract_response_text(response)
|
| 684 |
+
response_log: dict[str, Any] = {
|
| 685 |
+
"metadata": response_meta,
|
| 686 |
+
"output_text": text,
|
| 687 |
+
}
|
| 688 |
+
try:
|
| 689 |
+
parsed = _parse_json_object(text)
|
| 690 |
+
slots, background, validation_meta = _slots_from_layout_payload(
|
| 691 |
+
parsed,
|
| 692 |
+
canvas_w=canvas_w,
|
| 693 |
+
canvas_h=canvas_h,
|
| 694 |
+
chart_bbox=chart_bbox,
|
| 695 |
+
allow_chart_overlap=allow_chart_overlap,
|
| 696 |
+
chart_internal_slots=chart_internal_slots,
|
| 697 |
+
)
|
| 698 |
+
except Exception as exc:
|
| 699 |
+
response_log["parse_or_validation_error"] = f"{type(exc).__name__}: {exc}"
|
| 700 |
+
response_path.write_text(json.dumps(response_log, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 701 |
+
raise
|
| 702 |
+
|
| 703 |
+
response_log["parsed"] = parsed
|
| 704 |
+
response_log["validated"] = {
|
| 705 |
+
"background": background,
|
| 706 |
+
"reserved_slots": [asdict(slot) for slot in slots],
|
| 707 |
+
**validation_meta,
|
| 708 |
+
}
|
| 709 |
+
response_path.write_text(json.dumps(response_log, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 710 |
+
meta = {
|
| 711 |
+
"engine": "gpt",
|
| 712 |
+
"model": response_meta["model"],
|
| 713 |
+
"request_path": str(request_path),
|
| 714 |
+
"response_path": str(response_path),
|
| 715 |
+
**validation_meta,
|
| 716 |
+
}
|
| 717 |
+
return slots, background, meta
|
| 718 |
+
|
| 719 |
+
|
| 720 |
+
def _image_data_uri(path: Path) -> str:
|
| 721 |
+
payload = base64.b64encode(path.read_bytes()).decode("ascii")
|
| 722 |
+
return f"data:image/png;base64,{payload}"
|
| 723 |
+
|
| 724 |
+
|
| 725 |
+
def _slot_svg(slot: SlotPlanItem) -> str:
|
| 726 |
+
x, y, w, h = slot.bbox_px
|
| 727 |
+
cls = f"planned-slot planned-{slot.kind}-slot"
|
| 728 |
+
marker = html.escape(slot.id, quote=True)
|
| 729 |
+
element = html.escape(slot.element, quote=True)
|
| 730 |
+
anchor = html.escape(slot.anchor, quote=True)
|
| 731 |
+
collision = html.escape(slot.collision_rule, quote=True)
|
| 732 |
+
policy = html.escape(slot.asset_source_policy, quote=True)
|
| 733 |
+
return (
|
| 734 |
+
f'<g class="{cls}" data-asset-slot="{marker}" data-reserved-slot="{marker}" '
|
| 735 |
+
f'data-slot-kind="{html.escape(slot.kind, quote=True)}" '
|
| 736 |
+
f'data-slot-policy="{policy}" data-asset-source-policy="{policy}" '
|
| 737 |
+
f'data-bbox="{x},{y},{w},{h}" data-anchor="{anchor}" '
|
| 738 |
+
f'data-collision-rule="{collision}" data-planned-element="{element}" '
|
| 739 |
+
f'data-may-overlap-chart="{str(slot.may_overlap_chart).lower()}">'
|
| 740 |
+
f'<rect x="{x}" y="{y}" width="{w}" height="{h}" rx="10" '
|
| 741 |
+
f'fill="#E8EEF6" fill-opacity="0.62" stroke="#65758B" '
|
| 742 |
+
f'stroke-width="2" stroke-dasharray="10 8"/>'
|
| 743 |
+
"</g>"
|
| 744 |
+
)
|
| 745 |
+
|
| 746 |
+
|
| 747 |
+
def _write_planned_svg(
|
| 748 |
+
output_svg: Path,
|
| 749 |
+
chart_png: Path,
|
| 750 |
+
canvas_size: tuple[int, int],
|
| 751 |
+
chart_bbox: tuple[int, int, int, int],
|
| 752 |
+
slots: list[SlotPlanItem],
|
| 753 |
+
background: str,
|
| 754 |
+
) -> None:
|
| 755 |
+
canvas_w, canvas_h = canvas_size
|
| 756 |
+
chart_x, chart_y, chart_w, chart_h = chart_bbox
|
| 757 |
+
chart_href = _image_data_uri(chart_png)
|
| 758 |
+
slot_markup = "\n".join(_slot_svg(slot) for slot in slots)
|
| 759 |
+
svg = f'''<svg xmlns="http://www.w3.org/2000/svg" width="{canvas_w}" height="{canvas_h}" viewBox="0 0 {canvas_w} {canvas_h}" data-planned-slot-canvas="true">
|
| 760 |
+
<rect x="0" y="0" width="{canvas_w}" height="{canvas_h}" fill="{html.escape(background, quote=True)}"/>
|
| 761 |
+
<image class="sanitized-chart-raster" data-chart-source="sanitized-template-render" href="{chart_href}" x="{chart_x}" y="{chart_y}" width="{chart_w}" height="{chart_h}" preserveAspectRatio="xMidYMid meet"/>
|
| 762 |
+
{slot_markup}
|
| 763 |
+
</svg>'''
|
| 764 |
+
output_svg.parent.mkdir(parents=True, exist_ok=True)
|
| 765 |
+
output_svg.write_text(svg, encoding="utf-8")
|
| 766 |
+
|
| 767 |
+
|
| 768 |
+
def _render_planned_png(
|
| 769 |
+
output_png: Path,
|
| 770 |
+
chart_png: Path,
|
| 771 |
+
canvas_size: tuple[int, int],
|
| 772 |
+
chart_bbox: tuple[int, int, int, int],
|
| 773 |
+
slots: list[SlotPlanItem],
|
| 774 |
+
background_rgb: tuple[int, int, int] = (247, 248, 244),
|
| 775 |
+
) -> None:
|
| 776 |
+
canvas_w, canvas_h = canvas_size
|
| 777 |
+
output_png.parent.mkdir(parents=True, exist_ok=True)
|
| 778 |
+
canvas = Image.new("RGB", (canvas_w, canvas_h), background_rgb)
|
| 779 |
+
with Image.open(chart_png) as chart:
|
| 780 |
+
chart = chart.convert("RGBA")
|
| 781 |
+
x, y, w, h = chart_bbox
|
| 782 |
+
fitted = chart.copy()
|
| 783 |
+
fitted.thumbnail((w, h), Image.Resampling.LANCZOS)
|
| 784 |
+
paste_x = x + (w - fitted.width) // 2
|
| 785 |
+
paste_y = y + (h - fitted.height) // 2
|
| 786 |
+
canvas.paste(fitted.convert("RGB"), (paste_x, paste_y), fitted.getchannel("A"))
|
| 787 |
+
|
| 788 |
+
overlay = Image.new("RGBA", (canvas_w, canvas_h), (0, 0, 0, 0))
|
| 789 |
+
draw = ImageDraw.Draw(overlay)
|
| 790 |
+
colors = [
|
| 791 |
+
(93, 117, 150, 72),
|
| 792 |
+
(35, 150, 120, 68),
|
| 793 |
+
(208, 132, 48, 70),
|
| 794 |
+
(135, 88, 178, 70),
|
| 795 |
+
(60, 128, 190, 68),
|
| 796 |
+
]
|
| 797 |
+
for index, slot in enumerate(slots):
|
| 798 |
+
x, y, w, h = slot.bbox_px
|
| 799 |
+
fill = colors[index % len(colors)]
|
| 800 |
+
outline = (54, 66, 82, 155)
|
| 801 |
+
draw.rounded_rectangle((x, y, x + w, y + h), radius=10, fill=fill, outline=outline, width=2)
|
| 802 |
+
canvas = Image.alpha_composite(canvas.convert("RGBA"), overlay).convert("RGB")
|
| 803 |
+
canvas.save(output_png)
|
| 804 |
+
|
| 805 |
+
|
| 806 |
+
def _write_reference_map(
|
| 807 |
+
path: Path,
|
| 808 |
+
slots: list[SlotPlanItem],
|
| 809 |
+
chart_internal_slots: list[ProtectedRegion],
|
| 810 |
+
) -> None:
|
| 811 |
+
path.parent.mkdir(parents=True, exist_ok=True)
|
| 812 |
+
payload = {
|
| 813 |
+
"reserved_slots": [],
|
| 814 |
+
"chart_internal_slots": [
|
| 815 |
+
{
|
| 816 |
+
"id": region.id,
|
| 817 |
+
"reason": region.reason,
|
| 818 |
+
"bbox_px": list(region.bbox_px),
|
| 819 |
+
"source_slot": region.source_slot,
|
| 820 |
+
"anchor": region.anchor,
|
| 821 |
+
"collision_rule": region.collision_rule,
|
| 822 |
+
"semantic": region.semantic,
|
| 823 |
+
}
|
| 824 |
+
for region in chart_internal_slots
|
| 825 |
+
],
|
| 826 |
+
"protected_regions": [],
|
| 827 |
+
}
|
| 828 |
+
for slot in slots:
|
| 829 |
+
payload["reserved_slots"].append(
|
| 830 |
+
{
|
| 831 |
+
"id": slot.id,
|
| 832 |
+
"element": slot.element,
|
| 833 |
+
"reason": f"planned {slot.kind} slot for image-edit generation",
|
| 834 |
+
"slot_requirements": slot.slot_requirements,
|
| 835 |
+
"semantic_slot": {
|
| 836 |
+
"class_or_data_marker": f'data-asset-slot="{slot.id}"',
|
| 837 |
+
"anchor": slot.anchor,
|
| 838 |
+
"collision_rule": slot.collision_rule,
|
| 839 |
+
"asset_source_policy": slot.asset_source_policy,
|
| 840 |
+
"may_overlap_chart": slot.may_overlap_chart,
|
| 841 |
+
},
|
| 842 |
+
}
|
| 843 |
+
)
|
| 844 |
+
path.write_text(json.dumps(payload, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 845 |
+
|
| 846 |
+
|
| 847 |
+
def _parse_bbox_text(text: str | None) -> tuple[float, float, float, float] | None:
|
| 848 |
+
if not text:
|
| 849 |
+
return None
|
| 850 |
+
try:
|
| 851 |
+
parts = [float(part) for part in str(text).replace(",", " ").split() if part]
|
| 852 |
+
except ValueError:
|
| 853 |
+
return None
|
| 854 |
+
if len(parts) != 4:
|
| 855 |
+
return None
|
| 856 |
+
x, y, w, h = parts
|
| 857 |
+
if w <= 0 or h <= 0:
|
| 858 |
+
return None
|
| 859 |
+
return x, y, w, h
|
| 860 |
+
|
| 861 |
+
|
| 862 |
+
def _bbox_with_matrix(
|
| 863 |
+
bbox: tuple[float, float, float, float],
|
| 864 |
+
matrix: tuple[float, float, float, float, float, float],
|
| 865 |
+
) -> tuple[float, float, float, float]:
|
| 866 |
+
x, y, w, h = bbox
|
| 867 |
+
a, b, c, d, e, f = matrix
|
| 868 |
+
points = [
|
| 869 |
+
(a * x + c * y + e, b * x + d * y + f),
|
| 870 |
+
(a * (x + w) + c * y + e, b * (x + w) + d * y + f),
|
| 871 |
+
(a * x + c * (y + h) + e, b * x + d * (y + h) + f),
|
| 872 |
+
(a * (x + w) + c * (y + h) + e, b * (x + w) + d * (y + h) + f),
|
| 873 |
+
]
|
| 874 |
+
xs = [point[0] for point in points]
|
| 875 |
+
ys = [point[1] for point in points]
|
| 876 |
+
return min(xs), min(ys), max(xs) - min(xs), max(ys) - min(ys)
|
| 877 |
+
|
| 878 |
+
|
| 879 |
+
def _float_attr(elem: Any, name: str, default: float | None = None) -> float | None:
|
| 880 |
+
value = elem.get(name)
|
| 881 |
+
if value is None:
|
| 882 |
+
return default
|
| 883 |
+
try:
|
| 884 |
+
return float(str(value).strip())
|
| 885 |
+
except ValueError:
|
| 886 |
+
return default
|
| 887 |
+
|
| 888 |
+
|
| 889 |
+
def _element_local_bbox(elem: Any) -> tuple[float, float, float, float] | None:
|
| 890 |
+
tag = etree_local_name(elem)
|
| 891 |
+
if tag in {"rect", "image"}:
|
| 892 |
+
x = _float_attr(elem, "x", 0.0)
|
| 893 |
+
y = _float_attr(elem, "y", 0.0)
|
| 894 |
+
w = _float_attr(elem, "width")
|
| 895 |
+
h = _float_attr(elem, "height")
|
| 896 |
+
if x is not None and y is not None and w and h and w > 0 and h > 0:
|
| 897 |
+
return x, y, w, h
|
| 898 |
+
if tag == "circle":
|
| 899 |
+
cx = _float_attr(elem, "cx", 0.0)
|
| 900 |
+
cy = _float_attr(elem, "cy", 0.0)
|
| 901 |
+
r = _float_attr(elem, "r")
|
| 902 |
+
if cx is not None and cy is not None and r and r > 0:
|
| 903 |
+
return cx - r, cy - r, r * 2, r * 2
|
| 904 |
+
if tag == "ellipse":
|
| 905 |
+
cx = _float_attr(elem, "cx", 0.0)
|
| 906 |
+
cy = _float_attr(elem, "cy", 0.0)
|
| 907 |
+
rx = _float_attr(elem, "rx")
|
| 908 |
+
ry = _float_attr(elem, "ry")
|
| 909 |
+
if cx is not None and cy is not None and rx and ry and rx > 0 and ry > 0:
|
| 910 |
+
return cx - rx, cy - ry, rx * 2, ry * 2
|
| 911 |
+
return None
|
| 912 |
+
|
| 913 |
+
|
| 914 |
+
def etree_local_name(elem: Any) -> str:
|
| 915 |
+
try:
|
| 916 |
+
return elem.tag.rsplit("}", 1)[-1] if isinstance(elem.tag, str) else ""
|
| 917 |
+
except Exception:
|
| 918 |
+
return ""
|
| 919 |
+
|
| 920 |
+
|
| 921 |
+
def _collect_protected_regions(
|
| 922 |
+
chart_svg: Path,
|
| 923 |
+
source_chart_png: Path,
|
| 924 |
+
chart_bbox: tuple[int, int, int, int],
|
| 925 |
+
padding_px: int = 10,
|
| 926 |
+
) -> list[ProtectedRegion]:
|
| 927 |
+
try:
|
| 928 |
+
tree, root, svg_width, svg_height = _load_svg(chart_svg)
|
| 929 |
+
except Exception:
|
| 930 |
+
return []
|
| 931 |
+
del tree
|
| 932 |
+
with Image.open(source_chart_png) as chart_image:
|
| 933 |
+
png_w, png_h = chart_image.size
|
| 934 |
+
|
| 935 |
+
chart_x, chart_y, chart_w, chart_h = chart_bbox
|
| 936 |
+
svg_to_png_x = png_w / svg_width
|
| 937 |
+
svg_to_png_y = png_h / svg_height
|
| 938 |
+
png_to_canvas_x = chart_w / png_w
|
| 939 |
+
png_to_canvas_y = chart_h / png_h
|
| 940 |
+
nodes = root.xpath(
|
| 941 |
+
".//*[@data-asset-slot or @data-reserved-slot or @data-slot-id or @data-slot-kind]"
|
| 942 |
+
)
|
| 943 |
+
regions: list[ProtectedRegion] = []
|
| 944 |
+
for index, elem in enumerate(nodes, 1):
|
| 945 |
+
if not is_inside_chart_internal_slot(elem):
|
| 946 |
+
continue
|
| 947 |
+
raw_bbox = (
|
| 948 |
+
_parse_bbox_text(elem.get("data-bbox"))
|
| 949 |
+
or _parse_bbox_text(elem.get("data-asset-bbox"))
|
| 950 |
+
or _parse_bbox_text(elem.get("data-slot-bbox"))
|
| 951 |
+
)
|
| 952 |
+
if raw_bbox:
|
| 953 |
+
bbox_svg = _bbox_with_matrix(raw_bbox, _element_transform(elem, root))
|
| 954 |
+
else:
|
| 955 |
+
local_bbox = _element_local_bbox(elem)
|
| 956 |
+
if local_bbox is not None:
|
| 957 |
+
bbox_svg = _bbox_with_matrix(local_bbox, _element_transform(elem, root))
|
| 958 |
+
else:
|
| 959 |
+
abs_bbox = _absolute_bbox(elem, root)
|
| 960 |
+
if abs_bbox is None:
|
| 961 |
+
continue
|
| 962 |
+
bbox_svg = (abs_bbox.x, abs_bbox.y, abs_bbox.width, abs_bbox.height)
|
| 963 |
+
if bbox_svg is None:
|
| 964 |
+
continue
|
| 965 |
+
x, y, w, h = bbox_svg
|
| 966 |
+
if w <= 0 or h <= 0:
|
| 967 |
+
continue
|
| 968 |
+
px0 = chart_x + int(round(x * svg_to_png_x * png_to_canvas_x)) - padding_px
|
| 969 |
+
py0 = chart_y + int(round(y * svg_to_png_y * png_to_canvas_y)) - padding_px
|
| 970 |
+
px1 = chart_x + int(round((x + w) * svg_to_png_x * png_to_canvas_x)) + padding_px
|
| 971 |
+
py1 = chart_y + int(round((y + h) * svg_to_png_y * png_to_canvas_y)) + padding_px
|
| 972 |
+
px0 = max(0, px0)
|
| 973 |
+
py0 = max(0, py0)
|
| 974 |
+
px1 = min(chart_x + chart_w, px1)
|
| 975 |
+
py1 = min(chart_y + chart_h, py1)
|
| 976 |
+
if px1 <= px0 or py1 <= py0:
|
| 977 |
+
continue
|
| 978 |
+
slot_name = (
|
| 979 |
+
elem.get("data-asset-slot")
|
| 980 |
+
or elem.get("data-reserved-slot")
|
| 981 |
+
or elem.get("data-slot-id")
|
| 982 |
+
or elem.get("data-slot-kind")
|
| 983 |
+
or f"chart_internal_slot_{index}"
|
| 984 |
+
)
|
| 985 |
+
anchor = elem.get("data-asset-anchor") or elem.get("data-anchor") or elem.get("data-slot-anchor") or ""
|
| 986 |
+
collision_rule = elem.get("data-asset-collision") or elem.get("data-collision-rule") or ""
|
| 987 |
+
regions.append(
|
| 988 |
+
ProtectedRegion(
|
| 989 |
+
id=f"CHART_ICON_{len(regions) + 1:02d}",
|
| 990 |
+
reason="chart-internal icon/asset slot",
|
| 991 |
+
bbox_px=(px0, py0, px1 - px0, py1 - py0),
|
| 992 |
+
source_slot=str(slot_name),
|
| 993 |
+
anchor=str(anchor),
|
| 994 |
+
collision_rule=str(collision_rule),
|
| 995 |
+
semantic=(
|
| 996 |
+
"Generate or polish only the icon inside this chart slot; "
|
| 997 |
+
"preserve nearby chart labels, ranks, lines, points, and numeric values exactly."
|
| 998 |
+
),
|
| 999 |
+
)
|
| 1000 |
+
)
|
| 1001 |
+
return regions
|
| 1002 |
+
|
| 1003 |
+
|
| 1004 |
+
def build_planned_slot_package(
|
| 1005 |
+
chart_svg: Path,
|
| 1006 |
+
output_dir: Path,
|
| 1007 |
+
*,
|
| 1008 |
+
chart_png: Path | None = None,
|
| 1009 |
+
data_json: Path | None = None,
|
| 1010 |
+
canvas_width: int = 1536,
|
| 1011 |
+
canvas_height: int = 2048,
|
| 1012 |
+
allow_chart_overlap: bool = True,
|
| 1013 |
+
render_longest_side: int | None = None,
|
| 1014 |
+
layout_model: str | None = "gpt-5.5",
|
| 1015 |
+
layout_api_key: str | None = None,
|
| 1016 |
+
layout_api_key_env: str = "OPENAI_API_KEY",
|
| 1017 |
+
layout_base_url: str | None = None,
|
| 1018 |
+
layout_timeout_seconds: float = 120.0,
|
| 1019 |
+
layout_max_retries: int = 0,
|
| 1020 |
+
layout_max_output_tokens: int = 2500,
|
| 1021 |
+
deterministic_layout: bool = False,
|
| 1022 |
+
) -> PlannedSlotPackage:
|
| 1023 |
+
output_dir.mkdir(parents=True, exist_ok=True)
|
| 1024 |
+
sanitized_svg = output_dir / "sanitized_chart.svg"
|
| 1025 |
+
sanitized_png = output_dir / "sanitized_chart.png"
|
| 1026 |
+
sanitizer_report = output_dir / "sanitized_chart.report.json"
|
| 1027 |
+
sanitize_chart_svg(chart_svg, sanitized_svg, report_path=sanitizer_report)
|
| 1028 |
+
|
| 1029 |
+
if render_longest_side is not None:
|
| 1030 |
+
import os
|
| 1031 |
+
|
| 1032 |
+
previous = os.environ.get("RENDER_LONGEST_SIDE")
|
| 1033 |
+
os.environ["RENDER_LONGEST_SIDE"] = str(render_longest_side)
|
| 1034 |
+
try:
|
| 1035 |
+
render_svg_to_png(sanitized_svg, sanitized_png)
|
| 1036 |
+
finally:
|
| 1037 |
+
if previous is None:
|
| 1038 |
+
os.environ.pop("RENDER_LONGEST_SIDE", None)
|
| 1039 |
+
else:
|
| 1040 |
+
os.environ["RENDER_LONGEST_SIDE"] = previous
|
| 1041 |
+
else:
|
| 1042 |
+
render_svg_to_png(sanitized_svg, sanitized_png)
|
| 1043 |
+
|
| 1044 |
+
source_chart_png = sanitized_png if sanitized_png.is_file() else chart_png
|
| 1045 |
+
if source_chart_png is None or not source_chart_png.is_file():
|
| 1046 |
+
raise FileNotFoundError("planned slot package requires a chart PNG")
|
| 1047 |
+
|
| 1048 |
+
with Image.open(source_chart_png) as chart_image:
|
| 1049 |
+
chart_w, chart_h = chart_image.size
|
| 1050 |
+
max_w = int(canvas_width * 0.84)
|
| 1051 |
+
max_h = int(canvas_height * 0.58)
|
| 1052 |
+
scale = min(max_w / chart_w, max_h / chart_h, 1.0)
|
| 1053 |
+
placed_w = max(1, int(chart_w * scale))
|
| 1054 |
+
placed_h = max(1, int(chart_h * scale))
|
| 1055 |
+
chart_x = (canvas_width - placed_w) // 2
|
| 1056 |
+
chart_y = int(canvas_height * 0.30)
|
| 1057 |
+
chart_bbox = (chart_x, chart_y, placed_w, placed_h)
|
| 1058 |
+
|
| 1059 |
+
chart_internal_slots = _collect_protected_regions(chart_svg, source_chart_png, chart_bbox)
|
| 1060 |
+
data = _read_json(data_json)
|
| 1061 |
+
background = "#F7F8F4"
|
| 1062 |
+
if deterministic_layout or not layout_model:
|
| 1063 |
+
slots = _build_slots(
|
| 1064 |
+
canvas_width,
|
| 1065 |
+
canvas_height,
|
| 1066 |
+
chart_bbox,
|
| 1067 |
+
data,
|
| 1068 |
+
allow_chart_overlap,
|
| 1069 |
+
chart_internal_slots,
|
| 1070 |
+
)
|
| 1071 |
+
layout_engine: dict[str, Any] = {
|
| 1072 |
+
"engine": "deterministic",
|
| 1073 |
+
"model": None,
|
| 1074 |
+
"reason": "deterministic_layout option enabled" if deterministic_layout else "layout_model is empty",
|
| 1075 |
+
}
|
| 1076 |
+
else:
|
| 1077 |
+
slots, background, layout_engine = _build_gpt_slots(
|
| 1078 |
+
output_dir=output_dir,
|
| 1079 |
+
chart_png=source_chart_png,
|
| 1080 |
+
canvas_w=canvas_width,
|
| 1081 |
+
canvas_h=canvas_height,
|
| 1082 |
+
chart_bbox=chart_bbox,
|
| 1083 |
+
data=data,
|
| 1084 |
+
allow_chart_overlap=allow_chart_overlap,
|
| 1085 |
+
chart_internal_slots=chart_internal_slots,
|
| 1086 |
+
model=layout_model,
|
| 1087 |
+
api_key=layout_api_key,
|
| 1088 |
+
api_key_env=layout_api_key_env,
|
| 1089 |
+
base_url=layout_base_url,
|
| 1090 |
+
timeout_seconds=layout_timeout_seconds,
|
| 1091 |
+
max_retries=layout_max_retries,
|
| 1092 |
+
max_output_tokens=layout_max_output_tokens,
|
| 1093 |
+
)
|
| 1094 |
+
|
| 1095 |
+
slot_plan_path = output_dir / "slot_plan.json"
|
| 1096 |
+
planned_svg = output_dir / "planned_slots.svg"
|
| 1097 |
+
planned_png = output_dir / "planned_slots.png"
|
| 1098 |
+
reference_map = output_dir / "reference_element_map.json"
|
| 1099 |
+
slot_plan = {
|
| 1100 |
+
"canvas": {"width": canvas_width, "height": canvas_height, "background": background},
|
| 1101 |
+
"chart": {
|
| 1102 |
+
"sanitized_svg": str(sanitized_svg),
|
| 1103 |
+
"sanitized_png": str(source_chart_png),
|
| 1104 |
+
"bbox_px": list(chart_bbox),
|
| 1105 |
+
"overlap_policy": "editable_overlap_allowed" if allow_chart_overlap else "avoid_chart_overlap",
|
| 1106 |
+
},
|
| 1107 |
+
"layout_engine": layout_engine,
|
| 1108 |
+
"text_policy": "llm_generated",
|
| 1109 |
+
"chart_internal_slots": [asdict(region) for region in chart_internal_slots],
|
| 1110 |
+
"protected_regions": [],
|
| 1111 |
+
"reserved_slots": [asdict(slot) for slot in slots],
|
| 1112 |
+
}
|
| 1113 |
+
slot_plan_path.write_text(json.dumps(slot_plan, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 1114 |
+
_write_reference_map(reference_map, slots, chart_internal_slots)
|
| 1115 |
+
_write_planned_svg(planned_svg, source_chart_png, (canvas_width, canvas_height), chart_bbox, slots, background)
|
| 1116 |
+
_render_planned_png(planned_png, source_chart_png, (canvas_width, canvas_height), chart_bbox, slots)
|
| 1117 |
+
|
| 1118 |
+
return PlannedSlotPackage(
|
| 1119 |
+
slot_plan=slot_plan_path,
|
| 1120 |
+
planned_svg=planned_svg,
|
| 1121 |
+
planned_png=planned_png,
|
| 1122 |
+
reference_map=reference_map,
|
| 1123 |
+
sanitized_svg=sanitized_svg,
|
| 1124 |
+
sanitized_png=sanitized_png,
|
| 1125 |
+
sanitizer_report=sanitizer_report,
|
| 1126 |
+
)
|
| 1127 |
+
|
| 1128 |
+
|
| 1129 |
+
def parse_args(argv: Iterable[str] | None = None) -> argparse.Namespace:
|
| 1130 |
+
parser = argparse.ArgumentParser(description="Build a planned slot canvas from a chart-only SVG.")
|
| 1131 |
+
parser.add_argument("--chart-svg", type=Path, required=True)
|
| 1132 |
+
parser.add_argument("--chart-png", type=Path, default=None)
|
| 1133 |
+
parser.add_argument("--data-json", type=Path, default=None)
|
| 1134 |
+
parser.add_argument("--output-dir", type=Path, required=True)
|
| 1135 |
+
parser.add_argument("--canvas-width", type=int, default=1536)
|
| 1136 |
+
parser.add_argument("--canvas-height", type=int, default=2048)
|
| 1137 |
+
parser.add_argument("--disallow-chart-overlap", action="store_true")
|
| 1138 |
+
parser.add_argument("--render-longest-side", type=int, default=None)
|
| 1139 |
+
parser.add_argument("--layout-model", default="gpt-5.5")
|
| 1140 |
+
parser.add_argument("--layout-api-key-env", default="OPENAI_API_KEY")
|
| 1141 |
+
parser.add_argument("--layout-base-url", default=None)
|
| 1142 |
+
parser.add_argument("--layout-timeout", type=float, default=120.0)
|
| 1143 |
+
parser.add_argument("--layout-max-retries", type=int, default=0)
|
| 1144 |
+
parser.add_argument("--layout-max-output-tokens", type=int, default=2500)
|
| 1145 |
+
parser.add_argument(
|
| 1146 |
+
"--deterministic-layout",
|
| 1147 |
+
action="store_true",
|
| 1148 |
+
help="Use the old deterministic slot planner instead of GPT layout design.",
|
| 1149 |
+
)
|
| 1150 |
+
return parser.parse_args(argv)
|
| 1151 |
+
|
| 1152 |
+
|
| 1153 |
+
def main(argv: Iterable[str] | None = None) -> int:
|
| 1154 |
+
args = parse_args(argv)
|
| 1155 |
+
package = build_planned_slot_package(
|
| 1156 |
+
chart_svg=args.chart_svg,
|
| 1157 |
+
chart_png=args.chart_png,
|
| 1158 |
+
data_json=args.data_json,
|
| 1159 |
+
output_dir=args.output_dir,
|
| 1160 |
+
canvas_width=args.canvas_width,
|
| 1161 |
+
canvas_height=args.canvas_height,
|
| 1162 |
+
allow_chart_overlap=not args.disallow_chart_overlap,
|
| 1163 |
+
render_longest_side=args.render_longest_side,
|
| 1164 |
+
layout_model=args.layout_model,
|
| 1165 |
+
layout_api_key_env=args.layout_api_key_env,
|
| 1166 |
+
layout_base_url=args.layout_base_url,
|
| 1167 |
+
layout_timeout_seconds=args.layout_timeout,
|
| 1168 |
+
layout_max_retries=args.layout_max_retries,
|
| 1169 |
+
layout_max_output_tokens=args.layout_max_output_tokens,
|
| 1170 |
+
deterministic_layout=args.deterministic_layout,
|
| 1171 |
+
)
|
| 1172 |
+
print(json.dumps({k: str(v) for k, v in asdict(package).items()}, ensure_ascii=False))
|
| 1173 |
+
return 0
|
| 1174 |
+
|
| 1175 |
+
|
| 1176 |
+
if __name__ == "__main__":
|
| 1177 |
+
raise SystemExit(main())
|