feat: add site statistics

version: 0.2.5

README.md

@@ -14,3 +14,4 @@ After installation, the following commands will be available:
 
 -   `mccole lint`: internal project check
 -   `mccole render`: Markdown-to-HTML translator
+-   `mccole stats`: site statistics

mccole/clui.py

@@ -8,6 +8,7 @@
 
 from .lint import lint, parse_args as lint_parser
 from .render import render, parse_args as render_parser
+from .stats import stats, parse_args as stats_parser
 
 
 INSTALL_FILES = (
@@ -23,10 +24,13 @@ def main():
     '''Main driver.'''
     parser = argparse.ArgumentParser()
     parser.add_argument('--version', action='store_true', help='show version')
+
     subparsers = parser.add_subparsers(dest='cmd')
     install_parser(subparsers.add_parser('install', help='install files'))
     lint_parser(subparsers.add_parser('lint', help='check site'))
     render_parser(subparsers.add_parser('render', help='build site'))
+    stats_parser(subparsers.add_parser('stats', help='show stats'))
+
     opt = parser.parse_args()
     if opt.version:
         print(importlib.metadata.version('mccole'))
@@ -36,6 +40,8 @@ def main():
         lint(opt)
     elif opt.cmd == 'render':
         render(opt)
+    elif opt.cmd == 'stats':
+        stats(opt)
     else:
         print(f'unknown command {opt.cmd}', file=sys.stderr)
         sys.exit(1)

mccole/lint.py

@@ -6,15 +6,13 @@
 import re
 import tomli
 
-from .util import SUFFIXES, SUFFIXES_SRC, find_files
+from .util import MD_LINK_DEF, SUFFIXES, SUFFIXES_SRC, find_files, find_key_defs
 
 
 BIB_REF = re.compile(r"\[.+?\]\(b:(.+?)\)", re.MULTILINE)
 GLOSS_REF = re.compile(r"\[.+?\]\(g:(.+?)\)", re.MULTILINE)
-KEY_DEF = re.compile(r'^<span\s+id="(.+?)">.+?</span>\s*$', re.MULTILINE)
 MD_CODEBLOCK_FILE = re.compile(r"^```\s*\{\s*\.(.+?)\s+\#(.+?)\s*\}\s*$(.+?)^```\s*$", re.DOTALL + re.MULTILINE)
 MD_FILE_LINK = re.compile(r"\[(.+?)\]\((.+?)\)", re.MULTILINE)
-MD_LINK_DEF = re.compile(r"^\[(.+?)\]:\s+(.+?)\s*$", re.MULTILINE)
 MD_LINK_REF = re.compile(r"\[(.+?)\]\[(.+?)\]", re.MULTILINE)
 
 DEFAULT_CONFIG = {
@@ -83,15 +81,6 @@ def check_references(files, term, regexp, available):
     return ok
 
 
-def find_key_defs(files, term):
-    """Find key definitions in definition list file."""
-    candidates = [k for k in files if term in str(k).lower()]
-    if len(candidates) != 1:
-        return None
-    file_key = candidates[0]
-    return set(KEY_DEF.findall(files[file_key]))
-
-
 def is_missing(actual, available):
     """Is a file missing?"""
     return (not actual.exists()) or (

mccole/stats.py

@@ -0,0 +1,48 @@
+"""Report site stats."""
+
+import argparse
+from prettytable import PrettyTable
+
+from .util import MD_LINK_DEF, SUFFIXES, SUFFIXES_SRC, find_files, find_key_defs
+
+
+TABLE_FMT = {
+    "stat": "l",
+    "value": "r",
+}
+
+
+def stats(opt):
+    """Main driver."""
+    table = PrettyTable(field_names=TABLE_FMT.keys())
+    for k, v in TABLE_FMT.items():
+        table.align[k] = v
+
+    files = find_files(opt, set(["bin", opt.out]))
+    table.add_row(("bibliography entries", len(find_key_defs(files, "bibliography"))))
+    table.add_row(("glossary entries", len(find_key_defs(files, "glossary"))))
+    table.add_row(("link definitions", len(find_markdown_link_defs(files))))
+
+    print(table)
+
+
+def find_markdown_link_defs(files):
+    """Collect Markdown link key definnitions."""
+    found = set()
+    for filepath, content in files.items():
+        if filepath.suffix == ".md":
+            for link in MD_LINK_DEF.finditer(content):
+                found.add(link[0])
+    return found
+
+
+def parse_args(parser):
+    """Parse command-line arguments."""
+    parser.add_argument("--out", type=str, default="docs", help="output directory")
+    parser.add_argument("--root", type=str, default=".", help="root directory")
+    return parser
+
+
+if __name__ == "__main__":
+    opt = parse_args(argparse.ArgumentParser()).parse_args()
+    stats(opt)

mccole/util.py

@@ -1,8 +1,11 @@
 """Utilities."""
 
 from pathlib import Path
+import re
 
 
+KEY_DEF = re.compile(r'^<span\s+id="(.+?)">.+?</span>\s*$', re.MULTILINE)
+MD_LINK_DEF = re.compile(r"^\[(.+?)\]:\s+(.+?)\s*$", re.MULTILINE)
 SUFFIXES_BIN = {".ico", ".jpg", ".png"}
 SUFFIXES_SRC = {".css", ".html", ".js", ".md", ".py", ".sh", ".txt"}
 SUFFIXES_TXT = SUFFIXES_SRC | {".csv", ".json", ".svg"}
@@ -18,6 +21,15 @@ def find_files(opt, root_skips=[]):
     }
 
 
+def find_key_defs(files, term):
+    """Find key definitions in definition list file."""
+    candidates = [k for k in files if term in str(k).lower()]
+    if len(candidates) != 1:
+        return None
+    file_key = candidates[0]
+    return set(KEY_DEF.findall(files[file_key]))
+
+
 def find_symlinks(opt, root_skips=[]):
     """Collect all interesting files."""
     return [

pyproject.toml

@@ -10,7 +10,7 @@ mypkg = ["*.css", "*.js", "*.html"]
 
 [project]
 name = "mccole"
-version = "0.2.4"
+version = "0.2.5"
 authors = [
     {name = "Greg Wilson", email = "[email protected]"},
 ]
@@ -31,6 +31,7 @@ dependencies = [
     "html5validator>=0.4.2",
     "jinja2>=3.1.4",
     "markdown>=3.6",
+    "prettytable>=3.11",
     "ruff>=0.6.0",
     "tomli>=2.0.1",
 ]