rebased on top of #89

src/sphinx_tags/__init__.py

@@ -3,7 +3,6 @@
 """
 import os
 import re
-from collections import defaultdict
 from fnmatch import fnmatch
 from pathlib import Path
 from typing import List
@@ -32,29 +31,26 @@ class TagLinks(SphinxDirective):
     # Sphinx directive class attributes
     required_arguments = 0
     optional_arguments = 200  # Arbitrary.
-    final_argument_whitespace = True
     has_content = True
-
+    final_argument_whitespace = True
     # Custom attributes
     separator = ","
 
     def run(self):
-
-        # Undo splitting args by whitespace, and use our own separator (to support tags with spaces)
         if not (self.arguments or self.content):
             raise ExtensionError(
                 "Tags must be passed in either as an argument or in the body of the tag."
             )
-        tagline = []
 
+        tagline = []
         if self.arguments:
-            tagline.extend((" ".join(self.arguments)).split(self.separator))
-
+            tagline.extend(self.arguments)
         if self.content:
-            tagline.extend((" ".join(self.content)).split(self.separator))
+            # normalize white space and remove "\n"
+            tagline.extend((" ".join(self.content)).strip().split())
+
+        tags = [tag.strip() for tag in (" ".join(tagline)).split(self.separator)]
 
-        tags = [tag.strip() for tag in tagline]
-
         tag_dir = Path(self.env.app.srcdir) / self.env.app.config.tags_output_dir
         result = nodes.paragraph()
         result["classes"] = ["tags"]
@@ -63,7 +59,6 @@ def run(self):
 
         current_doc_dir = Path(self.env.doc2path(self.env.docname)).parent
         relative_tag_dir = Path(os.path.relpath(tag_dir, current_doc_dir))
-
         for tag in tags:
             count += 1
             # We want the link to be the path to the _tags folder, relative to
@@ -75,8 +70,6 @@ def run(self):
             #   |
             #    - current_doc_path
 
-            file_basename = _normalize_tag(tag)
-
             file_basename = _normalize_tag(tag)
             print(f"file_base: {file_basename} tag: {tag}")
             if self.env.app.config.tags_create_badges:
@@ -179,7 +172,10 @@ def create_file(
 
         """
         # Get sorted file paths for tag pages, relative to /docs/_tags
-        tag_page_paths = sorted(i.relpath(srcdir) for i in items)
+        tag_page_paths = sorted(
+            [i.relpath(srcdir) for i in items], key=lambda p: Path(p).stem
+        )
+        print(f"{tag_page_paths}")
 
         content = []
         if "md" in extension:
@@ -299,17 +295,16 @@ def assign_entries(app):
     pages = []
     tags = {}
 
-    # Get document paths in the project that match specified file extensions
-    doc_paths = get_matching_files(
-        app.srcdir,
-        include_patterns=[f"**.{extension}" for extension in app.config.tags_extension],
-        exclude_patterns=app.config.exclude_patterns,
-    )
-
-    for path in doc_paths:
-        entry = Entry(Path(app.srcdir) / path)
+    for docname in app.env.found_docs:
+        doctags = app.env.metadata[docname].get("tags", None)
+        if doctags is None:
+            continue  # skip if no tags
+        print(f"{docname}: {doctags}")
+        entry = Entry(app.env.doc2path(docname), doctags)
         entry.assign_to_tags(tags)
         pages.append(entry)
+
+    print(f"entry tags: {tags}, {pages}")
     return tags, pages
 
 
@@ -378,9 +373,6 @@ def setup(app):
         "html",
     )
 
-    # reset Entry list cache:
-    Entry.page_tags = defaultdict(list)
-
     # Update tags
     # TODO: tags should be updated after sphinx-gallery is generated, and the
     # gallery is also connected to builder-inited. Are there situations when

test/conftest.py

@@ -47,9 +47,9 @@ def create_symlinks():
     symlink_dest_dir = SOURCE_ROOT_DIR / "test-symlink"
 
     for file in SOURCE_ROOT_DIR.glob("test-rst/*.rst"):
-        (symlink_dest_dir / file.name).symlink_to(file)
+        symlink(file, symlink_dest_dir / file.name)
     for dir in SOURCE_ROOT_DIR.glob("test-rst/subdir*"):
-        (symlink_dest_dir / dir.name).symlink_to(dir, target_is_directory=True)
+        symlink(dir, symlink_dest_dir / dir.name, target_is_directory=True)
 
     yield
 

test/outputs/general/_tags/tagsindex.txt

@@ -5,11 +5,7 @@ Tags overview
 Tags
 ^^^^
 
-<<<<<<< HEAD
-* [{(tag 4)}] (1)
-=======
-* [{(tag   4)}] (2)
->>>>>>> 0afce42 (doc: allowing tags in body)
+* [{(tag 4)}] (2)
 
 * tag 3 (3)
 

test/outputs/general/page_5.txt

@@ -1,4 +1,4 @@
 Page 5
 ******
 
-Tags: tag_1, tag_5, tag2, tag 3, [{(tag   4)}]
+Tags: tag_1, tag_5, tag2, tag 3, [{(tag 4)}]

test/test_general_tags.py

@@ -31,7 +31,7 @@ def run_all_formats():
 
 @run_all_formats()
 def test_build(app: SphinxTestApp, status: StringIO, warning: StringIO):
-    app.build()
+    app.build(force_all=True)
     assert "build succeeded" in status.getvalue()
 
     # Build with notebooks and text output results in spurious errors "File Not Found: _tags/*.html"
@@ -43,37 +43,39 @@ def test_build(app: SphinxTestApp, status: StringIO, warning: StringIO):
 @pytest.mark.sphinx(confoverrides={"tags_create_tags": True})
 @run_all_formats()
 def test_index(app: SphinxTestApp):
-    app.build()
+    app.build(force_all=True)
     build_dir = Path(app.srcdir) / "_build" / "text"
     print(app.srcdir)
     # Check tags index page
-    contents = (build_dir / "_tags" / "tagsindex.txt").read_text()
-    expected_contents = (OUTPUT_DIR / "_tags" / "tagsindex.txt").read_text()
-
-    assert contents == expected_contents
+    contents = build_dir / "_tags" / "tagsindex.txt"
+    expected_contents = OUTPUT_DIR / "_tags" / "tagsindex.txt"
+    with open(contents, "r") as actual, open(expected_contents, "r") as expected:
+        assert list(actual) == list(expected)
 
     # Check full toctree created by index
-    contents = (build_dir / "index.txt").read_text()
-    expected_contents = (OUTPUT_DIR / "index.txt").read_text()
-    assert contents == expected_contents
+    contents = build_dir / "index.txt"
+    expected_contents = OUTPUT_DIR / "index.txt"
+    with open(contents, "r") as actual, open(expected_contents, "r") as expected:
+        assert list(actual) == list(expected)
 
 
 @pytest.mark.sphinx(confoverrides={"tags_create_tags": True})
 @run_all_formats()
 def test_tag_pages(app: SphinxTestApp):
-    app.build()
+    app.build(force_all=True)
     build_dir = Path(app.srcdir) / "_build" / "text"
 
     # Check all expected tag pages
     for tag in ["tag_1", "tag2", "tag-3", "tag-4", "tag_5", "test-tag-please-ignore"]:
-        contents = (build_dir / "_tags" / f"{tag}.txt").read_text()
-        expected_contents = (OUTPUT_DIR / "_tags" / f"{tag}.txt").read_text()
-        assert contents == expected_contents
+        contents = build_dir / "_tags" / f"{tag}.txt"
+        expected_contents = OUTPUT_DIR / "_tags" / f"{tag}.txt"
+        with open(contents, "r") as actual, open(expected_contents, "r") as expected:
+            assert list(actual) == list(expected)
 
 
 @run_all_formats()
 def test_tagged_pages(app: SphinxTestApp):
-    app.build()
+    app.build(force_all=True)
     build_dir = Path(app.srcdir) / "_build" / "text"
 
     # Check all expected tag pages
@@ -83,9 +85,10 @@ def test_tagged_pages(app: SphinxTestApp):
         Path("page_5.txt"),
         Path("subdir") / "page_3.txt",
     ]:
-        contents = (build_dir / page).read_text()
-        expected_contents = (OUTPUT_DIR / page).read_text()
-        assert contents == expected_contents
+        contents = build_dir / page
+        expected_contents = OUTPUT_DIR / page
+        with open(contents, "r") as actual, open(expected_contents, "r") as expected:
+            assert list(actual) == list(expected)
 
 
 def test_empty_taglinks():