stats: Add histogram generation to the pull command

* --generate-histogram command line option
* PULL_GENERATE_HISTOGRAM if you don't want to use the command line

Signed-off-by: Alek Ratzloff <alekratz@gmail.com>

chanbans/__main__.py

@@ -27,17 +27,25 @@ def parse_args():
         subcommands |= {subcommand}
         return subparsers.add_parser(subcommand, *args, **kwargs)
 
-    _pull_parser = add_subcommand(
+    pull_parser = add_subcommand(
         "pull",
         help="Pull bans from 4chan, save thumbnails, update the database, and exit",
     )
+    pull_parser.add_argument(
+        "--generate-histogram",
+        action="store_true",
+        help="Generate a histogram SVG if any new posts are pulled",
+        default=config.PULL_GENERATE_HISTOGRAM,
+    )
+    # TODO(args) --histogram-path
 
     _histogram_parser = add_subcommand(
         "hist",
         help="Generate histogram file.",
         # This doesn't work as expected for some reason. Doesn't get parsed correctly
-        #aliases=["histogram"],
+        # aliases=["histogram"],
     )
+    # TODO(args) --histogram-path
 
     _serve_parser = add_subcommand("serve", help="Start HTTP server")
     _help_parser = add_subcommand("help", help="Show this help message")
@@ -63,7 +71,7 @@ def main():
     )
     match args.command:
         case "pull":
-            asyncio.run(pull())
+            asyncio.run(pull(generate_histogram=args.generate_histogram))
         case "serve":
             run_app()
         case "hist":

chanbans/config.py

@@ -32,6 +32,14 @@ def one_of_default(name: str, values: Sequence[str], default_value: str) -> str:
     return value
 
 
+def default_bool(name: str, default_value: bool) -> bool:
+    return one_of_default(
+        name,
+        ("yes", "true", "1", "no", "false", "0"),
+        str(default_value).lower(),
+    ) in ("yes", "true", "1")
+
+
 THUMBS_DIR = Path(default("THUMBS_DIR", "thumbs"))
 CACHE_DIR = Path(default("CACHE_DIR", "cache"))
 
@@ -43,9 +51,9 @@ HTTP_RESULTS_PER_PAGE = default("HTTP_RESULTS_PER_PAGE", 100)
 
 STATIC_HANDLER = one_of_default("STATIC_HANDLER", ("remote", "local"), "local")
 STATIC_LOCAL_PATH = Path(default("STATIC_LOCAL_PATH", "static"))
-STATIC_LOCAL_FOLLOW_SYMLINKS = one_of_default(
-    "STATIC_LOCAL_FOLLOW_SYMLINKS", ("yes", "true", "1", "no", "false", "0"), "true"
-) in ("yes", "true", "1")
+STATIC_LOCAL_FOLLOW_SYMLINKS = default_bool("STATIC_LOCAL_FOLLOW_SYMLINKS")
 STATIC_ROOT = default("STATIC_ROOT", "/static")
 
 HISTOGRAM_PATH = Path(default("HISTOGRAM_PATH", "static/histogram.svg"))
+
+PULL_GENERATE_HISTOGRAM = default_bool("PULL_GENERATE_HISTOGRAM", False)

chanbans/hist.py

@@ -1,6 +1,11 @@
+import logging
+from pathlib import Path
+
 from .db import get_db
 from .config import HISTOGRAM_PATH
-from pathlib import Path
+
+
+log = logging.getLogger(__name__)
 
 
 def histogram_svg():
@@ -31,5 +36,6 @@ def histogram_svg():
 
 
 def generate_histogram_svg(path: Path = HISTOGRAM_PATH):
+    log.info("Generating and writing histogram SVG to %s", path)
     svg = histogram_svg()
     path = path.write_text(svg)

chanbans/pull.py

@@ -12,6 +12,7 @@ from bs4 import BeautifulSoup as Soup
 from . import config
 from .db import get_db
 from .files import file_cache
+from .hist import generate_histogram_svg
 
 
 log = logging.getLogger(__name__)
@@ -46,10 +47,12 @@ async def get_thumb(thumb_path: Union[str, Path], post: dict) -> Optional[bytes]
                 return content
 
 
-async def pull():
+async def pull(generate_histogram: bool = config.GENERATE_HISTOGRAM):
     # TODO(args) --db-path arg
     db = get_db()
 
+    updated = False
+
     # Get HTML
     html = await get_bans_html()
     # Get post JSON
@@ -121,6 +124,9 @@ async def pull():
                         post["now"],
                     )
 
+                # We are definitely updating this entry
+                updated = True
+
                 post["action"] = cols["action"]
                 # post['board'] = cols['board']
                 post["length"] = cols["length"]
@@ -202,5 +208,8 @@ async def pull():
             if "UNIQUE" not in msg:
                 log.exception("error inserting data")
                 log.info("Continuing")
+    if updated and generate_histogram:
+        generate_histogram_svg()
+
     # Finish off thumbnail jobs
     await asyncio.gather(*download_jobs)