From 8efb4c6f387d0993da35f501558b8f68908e1636 Mon Sep 17 00:00:00 2001
From: Tessa Walsh <tessa@bitarchivist.net>
Date: Mon, 28 Oct 2024 15:14:39 -0400
Subject: [PATCH] Parse full headers

---
 cdxj_indexer/postquery.py | 24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/cdxj_indexer/postquery.py b/cdxj_indexer/postquery.py
index 360f28b..2071af9 100644
--- a/cdxj_indexer/postquery.py
+++ b/cdxj_indexer/postquery.py
@@ -23,16 +23,16 @@ def append_method_query_from_req_resp(req, resp):
 
     url = req.rec_headers.get_header("WARC-Target-URI")
     method = req.http_headers.protocol
-    return append_method_query(method, content_type, len_, stream, url)
+    return append_method_query(method, content_type, len_, stream, url, req.http_headers)
 
 
 # ============================================================================
-def append_method_query(method, content_type, len_, stream, url):
+def append_method_query(method, content_type, len_, stream, url, request_headers):
     # if method == 'GET':
     #    return '', ''
 
     if method == "POST" or method == "PUT":
-        query = query_extract(content_type, len_, stream, url)
+        query = query_extract(content_type, len_, stream, url, request_headers)
     else:
         query = ""
 
@@ -49,7 +49,7 @@ def append_method_query(method, content_type, len_, stream, url):
 
 
 # ============================================================================
-def query_extract(mime, length, stream, url):
+def query_extract(mime, length, stream, url, request_headers):
     """
     Extract a url-encoded form POST/PUT from stream
     content length, return None
@@ -95,19 +95,19 @@ def handle_binary(query_data):
             query = handle_binary(query_data)
 
     elif mime.startswith("multipart/"):
-        env = {
-            "REQUEST_METHOD": "POST",
-            "CONTENT_TYPE": mime,
-            "CONTENT_LENGTH": len(query_data),
-        }
+        # env = {
+        #     "REQUEST_METHOD": "POST",
+        #     "CONTENT_TYPE": mime,
+        #     "CONTENT_LENGTH": len(query_data),
+        # }
 
-        args = dict(fp=BytesIO(query_data), environ=env, keep_blank_values=True)
+        # args = dict(fp=BytesIO(query_data), environ=env, keep_blank_values=True)
 
         try:
-            _, params = parse_options_header(mime)
+            _, params = parse_options_header(request_headers)
             boundary = params["boundary"]
             parser = MultipartParser(stream, boundary, charset="utf8")
-        except ValueError:
+        except KeyError:
             # Content-Type multipart/form-data may lack "boundary" info
             query = handle_binary(query_data)
         else: