From 45d0dd1ed5202d40b97b154072fc371c473d4e56 Mon Sep 17 00:00:00 2001
From: Gianluca Marraffa <gianluca.marraffa@sysdig.com>
Date: Wed, 31 Jan 2024 12:05:55 +0100
Subject: [PATCH] fix(mark2confluence): Check if mark headers are present at
 the beginning of the page (#18)

The function `has_mark_headers` was returning `True` when a mark header was present in the file.

However files with headers in the content, for example:

```markdown
# How to push to confluence

Create a markdown file with mark headers:
     <!-- Space: FOO ->
     # Page Title
```
Were also processed, causing the `mark` binary to fail because it couldn't find headers.

This PR fixes it by checking the presence of mark headers in the first line of the file.
---
 mark2confluence/main.py                             | 12 ++++++------
 .../with_mark_headers_in_the_content.md             |  5 +++++
 .../with_mark_parent_header.md                      |  0
 .../with_mark_space_header.md                       |  0
 .../with_mark_title_header.md                       |  0
 .../without_mark_headers.md                         |  0
 tests/test_main.py                                  | 13 +++++++------
 7 files changed, 18 insertions(+), 12 deletions(-)
 create mode 100644 tests/resources/markdown/test_begins_with_mark_headers/with_mark_headers_in_the_content.md
 rename tests/resources/markdown/{test_has_mark_headers => test_begins_with_mark_headers}/with_mark_parent_header.md (100%)
 rename tests/resources/markdown/{test_has_mark_headers => test_begins_with_mark_headers}/with_mark_space_header.md (100%)
 rename tests/resources/markdown/{test_has_mark_headers => test_begins_with_mark_headers}/with_mark_title_header.md (100%)
 rename tests/resources/markdown/{test_has_mark_headers => test_begins_with_mark_headers}/without_mark_headers.md (100%)

diff --git a/mark2confluence/main.py b/mark2confluence/main.py
index ad67c19..1b6b413 100755
--- a/mark2confluence/main.py
+++ b/mark2confluence/main.py
@@ -100,12 +100,12 @@ def publish(path: str)-> tuple:
   return True, None
 
 
-def has_mark_headers(path: str) -> bool:
-  space_re = re.compile("<!--.?(space|parent|title):.*-->", re.IGNORECASE)
+def begins_with_mark_headers(path: str, headers: List[str] = ["Space", "Parent", "Title"]) -> bool:
   with open(path, 'r+') as f:
-    data = f.read().split("\n")
-    for line in data:
-      if space_re.match(line):
+    first_row = f.readline()
+    for header in headers:
+      regex = re.compile(f"^<!--.?{header}:.*-->", re.IGNORECASE)
+      if regex.match(first_row):
         return True
   return False
 
@@ -270,7 +270,7 @@ def main()->int:
   default_parents = get_default_parents(cfg.inputs.DEFAULT_PARENTS)
   status = {}
   for path in files:
-    if path[-3:] == '.md' and has_mark_headers(path):
+    if path[-3:] == '.md' and begins_with_mark_headers(path):
       logger.info(f"Processing file {path}")
       inject_default_parents(path, default_parents)
 
diff --git a/tests/resources/markdown/test_begins_with_mark_headers/with_mark_headers_in_the_content.md b/tests/resources/markdown/test_begins_with_mark_headers/with_mark_headers_in_the_content.md
new file mode 100644
index 0000000..1e59103
--- /dev/null
+++ b/tests/resources/markdown/test_begins_with_mark_headers/with_mark_headers_in_the_content.md
@@ -0,0 +1,5 @@
+# I HAVE MARK HEADERS IN THE CONTENT
+
+```markdown
+<!-- Space: Content -->
+```
diff --git a/tests/resources/markdown/test_has_mark_headers/with_mark_parent_header.md b/tests/resources/markdown/test_begins_with_mark_headers/with_mark_parent_header.md
similarity index 100%
rename from tests/resources/markdown/test_has_mark_headers/with_mark_parent_header.md
rename to tests/resources/markdown/test_begins_with_mark_headers/with_mark_parent_header.md
diff --git a/tests/resources/markdown/test_has_mark_headers/with_mark_space_header.md b/tests/resources/markdown/test_begins_with_mark_headers/with_mark_space_header.md
similarity index 100%
rename from tests/resources/markdown/test_has_mark_headers/with_mark_space_header.md
rename to tests/resources/markdown/test_begins_with_mark_headers/with_mark_space_header.md
diff --git a/tests/resources/markdown/test_has_mark_headers/with_mark_title_header.md b/tests/resources/markdown/test_begins_with_mark_headers/with_mark_title_header.md
similarity index 100%
rename from tests/resources/markdown/test_has_mark_headers/with_mark_title_header.md
rename to tests/resources/markdown/test_begins_with_mark_headers/with_mark_title_header.md
diff --git a/tests/resources/markdown/test_has_mark_headers/without_mark_headers.md b/tests/resources/markdown/test_begins_with_mark_headers/without_mark_headers.md
similarity index 100%
rename from tests/resources/markdown/test_has_mark_headers/without_mark_headers.md
rename to tests/resources/markdown/test_begins_with_mark_headers/without_mark_headers.md
diff --git a/tests/test_main.py b/tests/test_main.py
index ddfeff4..1d1e7fb 100644
--- a/tests/test_main.py
+++ b/tests/test_main.py
@@ -32,12 +32,13 @@ def test_load_env_prefixes():
     assert main.cfg.actions["FOO"] == "foo"
     assert main.cfg.runner["FOO"] == "foo"
 
-def test_has_mark_headers():
-    resource_directory = f"{RESOURCE_DIR}/markdown/test_has_mark_headers"
-    assert main.has_mark_headers(f"{resource_directory}/with_mark_space_header.md")
-    assert main.has_mark_headers(f"{resource_directory}/with_mark_parent_header.md")
-    assert main.has_mark_headers(f"{resource_directory}/with_mark_title_header.md")
-    assert not main.has_mark_headers(f"{resource_directory}/without_mark_headers.md")
+def test_begins_with_mark_headers():
+    resource_directory = f"{RESOURCE_DIR}/markdown/test_begins_with_mark_headers"
+    assert main.begins_with_mark_headers(f"{resource_directory}/with_mark_space_header.md")
+    assert main.begins_with_mark_headers(f"{resource_directory}/with_mark_parent_header.md")
+    assert main.begins_with_mark_headers(f"{resource_directory}/with_mark_title_header.md")
+    assert not main.begins_with_mark_headers(f"{resource_directory}/without_mark_headers.md")
+    assert not main.begins_with_mark_headers(f"{resource_directory}/with_mark_headers_in_the_content.md")
 
 def test_check_header_template():
     assert main.check_header_template("Valid Jinja2 {{ var }}")