Upload extractor.py

agents/extractor.py

@@ -68,13 +68,16 @@ class TokenExtractor:
         self.spacing: dict[str, SpacingToken] = {}
         self.radius: dict[str, RadiusToken] = {}
         self.shadows: dict[str, ShadowToken] = {}
-        
+
+        # Foreground-background pairs extracted from actual DOM elements
+        self.fg_bg_pairs: list[dict] = []
+
         # CSS Variables collection
         self.css_variables: dict[str, str] = {}
-        
+
         # Font tracking
         self.font_families: dict[str, FontFamily] = {}
-        
+
         # Statistics
         self.total_elements = 0
         self.errors: list[str] = []
@@ -258,7 +261,70 @@ class TokenExtractor:
         """)
         
         return styles_data
-    
+
+    async def _extract_fg_bg_pairs(self, page: Page) -> list[dict]:
+        """
+        Extract actual foreground-background color pairs from visible DOM elements.
+
+        For each visible element that has a non-transparent text color, walk up the
+        ancestor chain to find the effective background color.  This gives us real
+        foreground/background pairs so we can do accurate WCAG AA checks instead of
+        only comparing every color against white/black.
+        """
+        pairs = await page.evaluate("""
+            () => {
+                const pairs = [];
+                const seen = new Set();
+
+                function rgbToHex(rgb) {
+                    if (!rgb || rgb === 'transparent' || rgb === 'rgba(0, 0, 0, 0)') return null;
+                    const match = rgb.match(/rgba?\\((\\d+),\\s*(\\d+),\\s*(\\d+)/);
+                    if (!match) return null;
+                    const r = parseInt(match[1]);
+                    const g = parseInt(match[2]);
+                    const b = parseInt(match[3]);
+                    return '#' + [r, g, b].map(c => c.toString(16).padStart(2, '0')).join('');
+                }
+
+                function getEffectiveBackground(el) {
+                    let current = el;
+                    while (current && current !== document.documentElement) {
+                        const bg = window.getComputedStyle(current).backgroundColor;
+                        if (bg && bg !== 'rgba(0, 0, 0, 0)' && bg !== 'transparent') {
+                            return rgbToHex(bg);
+                        }
+                        current = current.parentElement;
+                    }
+                    return '#ffffff';  // default page background
+                }
+
+                const elements = document.querySelectorAll('*');
+                elements.forEach(el => {
+                    const styles = window.getComputedStyle(el);
+                    if (styles.display === 'none' || styles.visibility === 'hidden') return;
+
+                    const fg = rgbToHex(styles.color);
+                    if (!fg) return;
+
+                    const bg = getEffectiveBackground(el);
+                    if (!bg) return;
+
+                    const key = fg + '|' + bg;
+                    if (seen.has(key)) return;
+                    seen.add(key);
+
+                    pairs.push({
+                        foreground: fg,
+                        background: bg,
+                        element: el.tagName.toLowerCase(),
+                    });
+                });
+
+                return pairs;
+            }
+        """)
+        return pairs or []
+
     async def _extract_css_variables(self, page: Page) -> dict:
         """
         Extract CSS custom properties (variables) from :root and stylesheets.
@@ -1038,7 +1104,14 @@ class TokenExtractor:
                         self._aggregate_colors(page_colors)
                     except Exception as e:
                         self.warnings.append(f"Page scan failed: {str(e)}")
-                    
+
+                    # 8. Extract foreground-background color pairs for real AA checks
+                    try:
+                        fg_bg = await self._extract_fg_bg_pairs(page)
+                        self.fg_bg_pairs.extend(fg_bg)
+                    except Exception as e:
+                        self.warnings.append(f"FG/BG pair extraction failed: {str(e)}")
+
                     # =========================================================
                     # Log extraction results for this page
                     # =========================================================