deploy: privacy-sanitizer backend 2026-05-05

app/services/extractor.py

@@ -1,10 +1,11 @@
 """Document text extraction service.
 
 Pure functions to extract plain text from uploaded documents.
-Supported formats: PDF, DOCX, TXT, MD.
+Supported formats: PDF, DOCX, TXT, MD, PPTX, CSV, HTML.
 """
 from __future__ import annotations
 
+import csv
 import io
 
 
@@ -78,6 +79,79 @@ def extract_plain(data: bytes) -> str:
     raise ExtractionError("Could not decode the file as text.")
 
 
+def extract_pptx(data: bytes) -> str:
+    try:
+        from pptx import Presentation
+    except ImportError as e:  # pragma: no cover
+        raise ExtractionError("PPTX support not installed on the server.") from e
+
+    try:
+        prs = Presentation(io.BytesIO(data))
+    except Exception as e:
+        raise ExtractionError("Could not parse PPTX — file may be corrupted.") from e
+
+    parts: list[str] = []
+    for slide in prs.slides:
+        for shape in slide.shapes:
+            if hasattr(shape, "text") and shape.text and shape.text.strip():
+                parts.append(shape.text.strip())
+
+    if not parts:
+        raise ExtractionError("No text could be extracted from this PPTX.")
+
+    return "\n".join(parts).strip()
+
+
+def extract_csv(data: bytes) -> str:
+    for encoding in ("utf-8", "utf-8-sig", "latin-1"):
+        try:
+            text = data.decode(encoding)
+            break
+        except UnicodeDecodeError:
+            continue
+    else:
+        raise ExtractionError("Could not decode the CSV file.")
+
+    try:
+        reader = csv.reader(io.StringIO(text))
+        rows = [" | ".join(cell.strip() for cell in row if cell.strip()) for row in reader]
+        result = "\n".join(r for r in rows if r)
+    except Exception as e:
+        raise ExtractionError("Could not parse CSV.") from e
+
+    if not result.strip():
+        raise ExtractionError("No text could be extracted from this CSV.")
+
+    return result.strip()
+
+
+def extract_html(data: bytes) -> str:
+    try:
+        from bs4 import BeautifulSoup
+    except ImportError as e:  # pragma: no cover
+        raise ExtractionError("HTML support not installed on the server.") from e
+
+    for encoding in ("utf-8", "utf-8-sig", "latin-1"):
+        try:
+            text = data.decode(encoding)
+            break
+        except UnicodeDecodeError:
+            continue
+    else:
+        raise ExtractionError("Could not decode the HTML file.")
+
+    soup = BeautifulSoup(text, "html.parser")
+    # Remove script and style elements
+    for tag in soup(["script", "style", "head", "meta", "link"]):
+        tag.decompose()
+
+    result = soup.get_text(separator="\n", strip=True)
+    if not result.strip():
+        raise ExtractionError("No text could be extracted from this HTML.")
+
+    return result.strip()
+
+
 # Mapping content-type / extension → handler
 _HANDLERS = {
     "pdf": extract_pdf,
@@ -85,6 +159,9 @@ _HANDLERS = {
     "txt": extract_plain,
     "md": extract_plain,
     "markdown": extract_plain,
+    "pptx": extract_pptx,
+    "csv": extract_csv,
+    "html": extract_html,
 }
 
 
@@ -98,14 +175,26 @@ def detect_kind(filename: str, content_type: str | None) -> str | None:
         return "md"
     if name.endswith(".txt"):
         return "txt"
+    if name.endswith(".pptx"):
+        return "pptx"
+    if name.endswith(".csv"):
+        return "csv"
+    if name.endswith(".html") or name.endswith(".htm"):
+        return "html"
 
     ct = (content_type or "").lower()
     if "pdf" in ct:
         return "pdf"
     if "wordprocessingml" in ct or "docx" in ct:
         return "docx"
+    if "presentationml" in ct or "pptx" in ct:
+        return "pptx"
     if "markdown" in ct:
         return "md"
+    if "csv" in ct:
+        return "csv"
+    if "html" in ct:
+        return "html"
     if ct.startswith("text/"):
         return "txt"
     return None
@@ -114,6 +203,6 @@ def detect_kind(filename: str, content_type: str | None) -> str | None:
 def extract_text(filename: str, content_type: str | None, data: bytes) -> str:
     kind = detect_kind(filename, content_type)
     if kind is None:
-        raise ExtractionError("Unsupported file type. Allowed: PDF, DOCX, TXT, MD.")
+        raise ExtractionError("Unsupported file type. Allowed: PDF, DOCX, TXT, MD, PPTX, CSV, HTML.")
     handler = _HANDLERS[kind]
     return handler(data)

requirements.txt

@@ -14,3 +14,5 @@ python-dateutil
 python-multipart==0.0.20
 pypdf==5.1.0
 python-docx==1.1.2
+python-pptx==1.0.2
+beautifulsoup4==4.13.4