Output HTML tables

Author: adelavega

ace/export.py

@@ -98,4 +98,18 @@ def export_database(db, foldername, skip_empty=True):
     }
 
     with (foldername / 'export.json').open('w') as f:
-        json.dump(export_md, f)
+        json.dump(export_md, f)
+
+    # Save table HTML files if available
+    tables_dir = foldername / 'tables'
+    tables_dir.mkdir(parents=True, exist_ok=True)
+    
+    for art in articles:
+        art_dir = tables_dir / str(art.id)
+        art_dir.mkdir(parents=True, exist_ok=True)
+        
+        for t in art.tables:
+            if t.input_html:
+                table_file = art_dir / f"{t.id}.html"
+                with table_file.open('w', encoding='utf-8') as f:
+                    f.write(t.input_html)

ace/ingest.py

@@ -49,13 +49,13 @@ def _parse_article(args):
             # Fallback to original source identification
             source = manager.identify_source(html)
             if source is None:
-                logger.warning("Could not identify source for %s" % f)
+                logger.info("Could not identify source for %s" % f)
                 return f, None
 
         article = source.parse_article(html, pmid, metadata_dir=metadata_dir, **kwargs)
         return f, article
     except Exception as e:
-        logger.warning("Error parsing article %s: %s" % (f, str(e)))
+        logger.info("Error parsing article %s: %s" % (f, str(e)))
         return f, None
 
 

ace/sources.py

@@ -1801,18 +1801,18 @@ def _extract_tables_from_javascript(self, soup):
                             # Break after finding and successfully parsing tables
                             break
                         else:
-                            logger.warning("No tables found in JavaScript data after parsing")
+                            logger.info("No tables found in JavaScript data after parsing")
                     else:
                         logger.debug("Could not find tfviewerdata assignment")
 
                 except Exception as e:
-                    logger.warning(f"Error extracting tables from JavaScript: {e}")
+                    logger.info(f"Error extracting tables from JavaScript: {e}")
                     import traceback
                     logger.debug(traceback.format_exc())
                     continue
 
         if not tables:
-            logger.warning("No tables could be extracted from JavaScript data")
+            logger.info("No tables could be extracted from JavaScript data")
 
         return tables