Update app.py

Chun updated to include header and footer in mapping list

app.py

@@ -17,7 +17,7 @@ FORMATTER_PATH = os.path.join(GLMOCR_BASE, "postprocess", "result_formatter.py")
 DEFAULT_ZONE_FRAC = float(os.environ.get("GLMOCR_DEFAULT_ZONE_FRAC", "0.08"))
 
 # ---------------------------------------------------------------------------
-# 1. Config: include headers/footers (remove from abandon lists)
+# 1. Config: include headers/footers (remove from abandon, add to text mapping)
 # ---------------------------------------------------------------------------
 with open(CONFIG_PATH, "r") as f:
     config = yaml.safe_load(f)
@@ -25,19 +25,33 @@ with open(CONFIG_PATH, "r") as f:
 config["pipeline"]["maas"]["enabled"] = True
 config["pipeline"]["maas"]["api_key"] = "4570c28bdea5493c9efae9dae68edc66.sGbA9DLlcX1GlvqV"
 
-to_remove = {"header", "footer"}
-for key in ("layout", "result_formatter"):
-    if key not in config.get("pipeline", {}):
-        continue
-    section = config["pipeline"][key]
-    if key == "layout" and "label_task_mapping" in section:
-        abandon = section["label_task_mapping"].get("abandon")
-        if isinstance(abandon, list):
-            section["label_task_mapping"]["abandon"] = [x for x in abandon if x not in to_remove]
-    elif key == "result_formatter" and "abandon" in section:
-        abandon = section["abandon"]
-        if isinstance(abandon, list):
-            section["abandon"] = [x for x in abandon if x not in to_remove]
+to_include_as_text = {"header", "footer"}
+layout_section = config.get("pipeline", {}).get("layout", {})
+if "label_task_mapping" in layout_section:
+    mapping = layout_section["label_task_mapping"]
+    # Remove header/footer from abandon so they are not discarded
+    abandon = mapping.get("abandon")
+    if isinstance(abandon, list):
+        mapping["abandon"] = [x for x in abandon if x not in to_include_as_text]
+    # Add header/footer to text so they are OCR'd and output as text
+    text_labels = mapping.get("text")
+    if isinstance(text_labels, list):
+        for label in to_include_as_text:
+            if label not in text_labels:
+                text_labels.append(label)
+
+formatter_section = config.get("pipeline", {}).get("result_formatter", {})
+if "abandon" in formatter_section:
+    abandon = formatter_section["abandon"]
+    if isinstance(abandon, list):
+        formatter_section["abandon"] = [x for x in abandon if x not in to_include_as_text]
+# Include header/footer in visualization/output text category
+if "label_visualization_mapping" in formatter_section:
+    text_vis = formatter_section["label_visualization_mapping"].get("text")
+    if isinstance(text_vis, list):
+        for label in to_include_as_text:
+            if label not in text_vis:
+                text_vis.append(label)
 
 with open(CONFIG_PATH, "w") as f:
     yaml.dump(config, f, default_flow_style=False, sort_keys=False)