From 3ebdd2960716b0af7d3daf40d880b3f786e3afe2 Mon Sep 17 00:00:00 2001
From: Levente Hunyadi <levente.hunyadi@instructure.com>
Date: Fri, 11 Oct 2024 01:16:10 +0200
Subject: [PATCH] Set title from front matter if present

---
 md2conf/application.py | 13 +++++++++++++
 md2conf/converter.py   | 26 +++++++++++++++-----------
 md2conf/mermaid.py     |  5 ++---
 requirements.txt       |  1 +
 setup.cfg              |  1 +
 5 files changed, 32 insertions(+), 14 deletions(-)

diff --git a/md2conf/application.py b/md2conf/application.py
index c91ed4e..dbf04fd 100644
--- a/md2conf/application.py
+++ b/md2conf/application.py
@@ -3,6 +3,8 @@
 from pathlib import Path
 from typing import Dict, List, Optional
 
+import yaml
+
 from .api import ConfluencePage, ConfluenceSession
 from .converter import (
     ConfluenceDocument,
@@ -10,6 +12,7 @@
     ConfluencePageMetadata,
     ConfluenceQualifiedID,
     attachment_name,
+    extract_frontmatter,
     extract_qualified_id,
     read_qualified_id,
 )
@@ -137,6 +140,8 @@ def _get_or_create_page(
             document = f.read()
 
         qualified_id, document = extract_qualified_id(document)
+        frontmatter, document = extract_frontmatter(document)
+
         if qualified_id is not None:
             confluence_page = self.api.get_page(
                 qualified_id.page_id, space_key=qualified_id.space_key
@@ -147,6 +152,14 @@ def _get_or_create_page(
                     f"expected: parent page ID for Markdown file with no linked Confluence page: {absolute_path}"
                 )
 
+            # assign title from frontmatter if present
+            if title is None and frontmatter is not None:
+                properties = yaml.safe_load(frontmatter)
+                if isinstance(properties, dict):
+                    property_title = properties.get("title")
+                    if isinstance(property_title, str):
+                        title = property_title
+
             confluence_page = self._create_page(
                 absolute_path, document, title, parent_id
             )
diff --git a/md2conf/converter.py b/md2conf/converter.py
index 5946a6a..f02675e 100644
--- a/md2conf/converter.py
+++ b/md2conf/converter.py
@@ -844,16 +844,16 @@ class DocumentError(RuntimeError):
     pass
 
 
-def extract_value(pattern: str, string: str) -> Tuple[Optional[str], str]:
+def extract_value(pattern: str, text: str) -> Tuple[Optional[str], str]:
     values: List[str] = []
 
     def _repl_func(matchobj: re.Match) -> str:
         values.append(matchobj.group(1))
         return ""
 
-    string = re.sub(pattern, _repl_func, string, 1, re.ASCII)
+    text = re.sub(pattern, _repl_func, text, 1, re.ASCII)
     value = values[0] if values else None
-    return value, string
+    return value, text
 
 
 @dataclass
@@ -866,20 +866,24 @@ def __init__(self, page_id: str, space_key: Optional[str] = None):
         self.space_key = space_key
 
 
-def extract_qualified_id(string: str) -> Tuple[Optional[ConfluenceQualifiedID], str]:
+def extract_qualified_id(text: str) -> Tuple[Optional[ConfluenceQualifiedID], str]:
     "Extracts the Confluence page ID and space key from a Markdown document."
 
-    page_id, string = extract_value(r"<!--\s+confluence-page-id:\s*(\d+)\s+-->", string)
+    page_id, text = extract_value(r"<!--\s+confluence-page-id:\s*(\d+)\s+-->", text)
 
     if page_id is None:
-        return None, string
+        return None, text
 
     # extract Confluence space key
-    space_key, string = extract_value(
-        r"<!--\s+confluence-space-key:\s*(\S+)\s+-->", string
-    )
+    space_key, text = extract_value(r"<!--\s+confluence-space-key:\s*(\S+)\s+-->", text)
+
+    return ConfluenceQualifiedID(page_id, space_key), text
+
+
+def extract_frontmatter(text: str) -> Tuple[Optional[str], str]:
+    "Extracts the front matter from a Markdown document."
 
-    return ConfluenceQualifiedID(page_id, space_key), string
+    return extract_value(r"(?ms)\A---$(.+?)^---$", text)
 
 
 def read_qualified_id(absolute_path: Path) -> Optional[ConfluenceQualifiedID]:
@@ -956,7 +960,7 @@ def __init__(
         )
 
         # extract frontmatter
-        frontmatter, text = extract_value(r"(?ms)\A---$(.+?)^---$", text)
+        frontmatter, text = extract_frontmatter(text)
 
         # convert to HTML
         html = markdown_to_html(text)
diff --git a/md2conf/mermaid.py b/md2conf/mermaid.py
index 63c447f..fc83da0 100644
--- a/md2conf/mermaid.py
+++ b/md2conf/mermaid.py
@@ -49,10 +49,9 @@ def render(source: str, output_format: Literal["png", "svg"] = "png") -> bytes:
         "--outputFormat",
         output_format,
     ]
+    root = os.path.dirname(__file__)
     if is_docker():
-        cmd.extend(
-            ["-p", os.path.join(os.path.dirname(__file__), "puppeteer-config.json")]
-        )
+        cmd.extend(["-p", os.path.join(root, "puppeteer-config.json")])
     LOGGER.debug(f"Executing: {' '.join(cmd)}")
     try:
         proc = subprocess.Popen(
diff --git a/requirements.txt b/requirements.txt
index fee4195..910a539 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,6 +4,7 @@ types-lxml
 markdown
 types-markdown
 pymdown-extensions
+pyyaml
 requests
 types-requests
 
diff --git a/setup.cfg b/setup.cfg
index da2f566..65da147 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -33,6 +33,7 @@ install_requires =
     markdown >= 3.6
     types-markdown >= 3.6
     pymdown-extensions >= 10.9
+    pyyaml >= 6.0
     requests >= 2.32
     types-requests >= 2.32