fixup! Mutualize getting ArcWarcRecord content

openzim · Jan 25, 2024 · a597f3a · a597f3a
1 parent 990ea36
commit a597f3a
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 6 deletions.
diff --git a/src/warc2zim/converter.py b/src/warc2zim/converter.py
@@ -47,7 +47,7 @@
 from warc2zim.items import StaticArticle, WARCPayloadItem
 from warc2zim.url_rewriting import FUZZY_RULES, normalize
 from warc2zim.utils import (
-    get_arc_warc_record_content,
+    get_record_content,
     get_record_mime_type,
     get_record_url,
     get_version,
@@ -327,7 +327,7 @@ def gather_information_from_warc(self):
                 main_page_found = True
                 continue
 
-            content = get_arc_warc_record_content(record)
+            content = get_record_content(record)
 
             if not self.title:
                 self.title = parse_title(content)
@@ -417,7 +417,7 @@ def retrieve_illustration(self):
                         "Illustration_48x48_at_1"
                     ]
                     return
-                self.illustration = get_arc_warc_record_content(record)
+                self.illustration = get_record_content(record)
 
         # favicon_url not in WARC ; downloading
         try:

diff --git a/src/warc2zim/items.py b/src/warc2zim/items.py
@@ -20,7 +20,7 @@
 from warc2zim.content_rewriting.js import JsRewriter
 from warc2zim.url_rewriting import ArticleUrlRewriter
 from warc2zim.utils import (
-    get_arc_warc_record_content,
+    get_record_content,
     get_record_mime_type,
     get_record_url,
 )
@@ -44,7 +44,7 @@ def __init__(
         self.path = path
         self.mimetype = get_record_mime_type(record)
         self.title = ""
-        self.content = get_arc_warc_record_content(record)
+        self.content = get_record_content(record)
 
         if getattr(record, "method", "GET") == "POST":
             return

diff --git a/src/warc2zim/utils.py b/src/warc2zim/utils.py
@@ -52,7 +52,7 @@ def to_string(input_: str | bytes) -> str:
     return input_  # pyright: ignore[reportGeneralTypeIssues, reportReturnType]
 
 
-def get_arc_warc_record_content(record: ArcWarcRecord):
+def get_record_content(record: ArcWarcRecord):
     if hasattr(record, "buffered_stream"):
         stream = (
             record.buffered_stream  # pyright: ignore [reportGeneralTypeIssues, reportAttributeAccessIssue]