doocus first ver

doocus/extractors/pdf.py

@@ -0,0 +1,73 @@
+"""
+PDF-family extractor.
+
+Text per page via pypdf, plus the document info dict (title/author/dates). When
+options['render'] is set and pdf2image (+ system poppler) is available, page 1 is
+rendered to thumb.jpg.
+"""
+import io
+from pathlib import Path
+
+from .base import ExtractionResult
+
+EXTRACTOR_ID = "pdf/pypdf@1"
+
+
+def _info_metadata(reader) -> dict:
+    out = {}
+    info = getattr(reader, "metadata", None)
+    if not info:
+        return out
+    for src, dst in (("title", "title"), ("author", "author"),
+                     ("creation_date", "created"), ("modification_date", "modified_doc")):
+        try:
+            val = getattr(info, src, None)
+        except Exception:
+            val = None
+        if val:
+            out[dst] = str(val)
+    return out
+
+
+def _render_first_page(source: Path) -> bytes:
+    from pdf2image import convert_from_path  # needs system poppler
+    images = convert_from_path(str(source), first_page=1, last_page=1, dpi=100)
+    if not images:
+        return b""
+    im = images[0]
+    im.thumbnail((640, 640))
+    buf = io.BytesIO()
+    im.convert("RGB").save(buf, format="JPEG", quality=80)
+    return buf.getvalue()
+
+
+def extract(source: Path, options: dict) -> ExtractionResult:
+    from pypdf import PdfReader  # lazy
+
+    warnings = []
+    reader = PdfReader(str(source))
+    pages = reader.pages
+    blocks = []
+    for i, page in enumerate(pages, 1):
+        try:
+            text = page.extract_text() or ""
+        except Exception as e:
+            text = ""
+            warnings.append(f"page {i} text extraction failed: {e}")
+        blocks.append(f"## Page {i}\n\n{text.strip()}")
+    content = "\n\n".join(blocks) + "\n"
+
+    metadata = _info_metadata(reader)
+    metadata["pages"] = len(pages)
+
+    thumb = None
+    if options.get("render"):
+        try:
+            thumb = _render_first_page(source) or None
+        except ImportError:
+            warnings.append("--render requested but pdf2image/poppler not available")
+        except Exception as e:
+            warnings.append(f"page render failed: {e}")
+
+    return ExtractionResult(content=content, metadata=metadata, thumb=thumb,
+                            warnings=warnings, extractor=EXTRACTOR_ID)