From 1b2d6aff6ccf72fdb292a1f05bb41bf9633a8f55 Mon Sep 17 00:00:00 2001
From: David Luevano Alvarado <david@luevano.xyz>
Date: Tue, 25 Apr 2023 04:03:48 -0600
Subject: refactor tests and add more typing
---
src/pyssg/configuration.py | 2 +
src/pyssg/md_parser.py | 12 ++++--
src/pyssg/page.py | 101 ++++++++++++++++++++-------------------------
src/pyssg/plt/default.yaml | 5 ---
4 files changed, 55 insertions(+), 65 deletions(-)
(limited to 'src')
diff --git a/src/pyssg/configuration.py b/src/pyssg/configuration.py
index e2dc26b..c7b4248 100644
--- a/src/pyssg/configuration.py
+++ b/src/pyssg/configuration.py
@@ -12,6 +12,7 @@ DEFAULT_CONFIG_PATH: str = '$XDG_CONFIG_HOME/pyssg/config.yaml'
VERSION: str = version('pyssg')
+# TODO: add checking for extensions config (such as pymdvar)
def __check_well_formed_config(config: dict[str, Any],
config_base: list[dict[str, Any]],
prefix_key: str = '') -> None:
@@ -80,6 +81,7 @@ def get_static_config(sc_package: str = 'static_config.yaml',
log.debug('reading and setting static config')
config: dict[str, Any] = get_parsed_yaml(sc_package, plt_resource)[0]
+ # TODO: move this to utils and update the tests
def __time(fmt: str) -> str:
return datetime.now(tz=timezone.utc).strftime(config['fmt'][fmt])
diff --git a/src/pyssg/md_parser.py b/src/pyssg/md_parser.py
index 3ef297d..3b62985 100644
--- a/src/pyssg/md_parser.py
+++ b/src/pyssg/md_parser.py
@@ -4,6 +4,7 @@ from logging import Logger, getLogger
from markdown import Markdown
from yafg import YafgExtension
+from pymdvar import VariableExtension
from markdown_checklist.extension import ChecklistExtension
from .database import Database
@@ -12,13 +13,17 @@ from .page import Page
log: Logger = getLogger(__name__)
-def _get_md_obj() -> Markdown:
+# TODO: add configuration testing for extensions config (pymdvar for ex)
+def get_md_obj(variables: dict[str, str] = dict(),
+ enable_env: bool = False) -> Markdown:
exts: list = ['extra',
'meta',
'sane_lists',
'smarty',
'toc',
'wikilinks',
+ VariableExtension(variables=variables,
+ enable_env=enable_env),
# stripTitle generates an error when True,
# if there is no title attr
YafgExtension(stripTitle=False,
@@ -50,7 +55,7 @@ class MDParser:
self.config: dict = config
self.dir_config: dict = dir_config
self.db: Database = db
- self.md: Markdown = _get_md_obj()
+ self.md: Markdown = get_md_obj()
self.all_files: list[Page] = []
self.all_tags: list[tuple[str, str]] = []
@@ -65,7 +70,8 @@ class MDParser:
log.debug('parsing md into html')
content: str = self.md.reset().convert(open(src_file).read())
- # ignoring md.Meta type as it is not yet defined (because it is from an extension)
+ # ignoring md.Meta type as it is not yet defined
+ # (because it is from an extension)
page: Page = Page(f,
self.db.e[f].ctimestamp,
self.db.e[f].mtimestamp,
diff --git a/src/pyssg/page.py b/src/pyssg/page.py
index 6b8916d..4902bea 100644
--- a/src/pyssg/page.py
+++ b/src/pyssg/page.py
@@ -1,22 +1,21 @@
-import sys
from datetime import datetime, timezone
from logging import Logger, getLogger
+from typing import Any
log: Logger = getLogger(__name__)
class Page:
- def __init__(self,
- name: str,
- ctime: float,
- mtime: float,
- html: str,
- toc: str,
- toc_tokens: list[str],
- meta: dict,
- config: dict,
- dir_config: dict) -> None:
- log.debug('initializing the page object with name "%s"', name)
+ def __init__(self, name: str,
+ ctime: float,
+ mtime: float,
+ html: str,
+ toc: str,
+ toc_tokens: list[str],
+ meta: dict[str, Any],
+ config: dict[str, Any],
+ dir_config: dict[str, Any]) -> None:
+ log.debug('initializing a page object with name "%s"', name)
# initial data
self.name: str = name
self.ctimestamp: float = ctime
@@ -24,9 +23,9 @@ class Page:
self.content: str = html
self.toc: str = toc
self.toc_tokens: list[str] = toc_tokens
- self.meta: dict = meta
- self.config: dict = config
- self.dir_config: dict = dir_config
+ self.meta: dict[str, Any] = meta
+ self.config: dict[str, Any] = config
+ self.dir_config: dict[str, Any] = dir_config
# data from self.meta
self.title: str
@@ -48,25 +47,25 @@ class Page:
self.next: Page | None = None
self.previous: Page | None = None
- # also from self.meta, but for og metadata
- self.og: dict[str, str] = dict()
-
def __lt__(self, other):
return self.ctimestamp < other.ctimestamp
- def __get_meta(self, var: str, or_else: str | list[str]) -> str | list[str]:
+ def __get_meta(self, var: str,
+ or_else: str | list[str] = '') -> str | list[str] | Any:
if var in self.meta:
log.debug('getting metadata "%s"', var)
return self.meta[var]
else:
- log.debug('getting metadata "%s" failed, using optional value "%s"', var, or_else)
+ log.debug('getting metadata "%s" failed, using optional value "%s"',
+ var, or_else)
return or_else
def cdate(self, format: str) -> str:
if format in self.config['fmt']:
return self.cdatetime.strftime(self.config['fmt'][format])
else:
- log.warning('format "%s" not found in config["fmt"], returning empty string', format)
+ log.warning('format "%s" not found in config, returning '
+ 'empty string', format)
return ''
def mdate(self, format: str) -> str:
@@ -74,28 +73,32 @@ class Page:
log.warning('no mdatetime found, can\'t return a formatted string')
return ''
if format in self.config['fmt']:
- return self.mdatetime.strftime(self.config['fmt'][format]) # type: ignore
+ return self.mdatetime.strftime(self.config['fmt'][format])
else:
- log.warning('format "%s" not found in config["fmt"], returning empty string', format)
+ log.warning('format "%s" not found in config, returning '
+ 'empty string', format)
return ''
+ def from_timestamp(self, timestamp: float) -> datetime:
+ return datetime.fromtimestamp(timestamp, tz=timezone.utc)
+
# parses meta from self.meta, for og, it prioritizes,
# the actual og meta
def parse_metadata(self):
log.debug('parsing metadata for file "%s"', self.name)
- self.title = self.__get_meta('title', [''])[0]
+ self.title = str(self.__get_meta('title'))
self.author = list(self.__get_meta('author', ['']))
- self.summary = self.__get_meta('summary', [''])[0]
- self.lang = self.__get_meta('lang', ['en'])[0]
+ self.summary = str(self.__get_meta('summary'))
+ self.lang = str(self.__get_meta('lang', 'en'))
log.debug('parsing timestamp')
- self.cdatetime = datetime.fromtimestamp(self.ctimestamp, tz=timezone.utc)
+ self.cdatetime = self.from_timestamp(self.ctimestamp)
self.cdate_rss = self.cdate('rss_date')
self.cdate_sitemap = self.cdate('sitemap_date')
if self.mtimestamp != 0.0:
log.debug('parsing modified timestamp')
- self.mdatetime = datetime.fromtimestamp(self.mtimestamp, tz=timezone.utc)
+ self.mdatetime = self.from_timestamp(self.mtimestamp)
self.mdate_rss = self.mdate('rss_date')
self.mdate_sitemap = self.mdate('sitemap_date')
else:
@@ -108,30 +111,35 @@ class Page:
tags_only.sort()
for t in tags_only:
- # need to specify dir_config['url'] as it is a hardcoded tag url
- self.tags.append((t, f'{self.dir_config["url"]}/tag/@{t}.html'))
+ # need to specify dir_config['url'] as it is
+ # a hardcoded tag url
+ tag_url: str = f'{self.dir_config["url"]}/tag/@{t}.html'
+ self.tags.append((t, tag_url))
else:
log.debug('no tags to parse')
- log.debug('parsing url')
- # no need to specify dir_config['url'] as self.name already contains the relative url
- self.url = f'{self.config["url"]["main"]}/{self.name.replace(".md", ".html")}'
+ log.debug('parsing page url')
+ # no need to specify dir_config['url'] as self.name already
+ # contains the relative url
+ name_html: str = self.name.replace(".md", ".html")
+ self.url = f'{self.config["url"]["main"]}/{name_html}'
log.debug('final url "%s"', self.url)
log.debug('parsing image url')
default_image_url: str = ''
if 'default_image' in self.config['url']:
- log.debug('"default_image" url found, will use if no "image_url" is found')
+ log.debug('"default_image" url found, will use if no "image_url" '
+ 'is found')
default_image_url = self.config['url']['default_image']
image_url: str
- image_url = self.__get_meta('image_url', [default_image_url])[0]
+ image_url = str(self.__get_meta('image_url', default_image_url))
if image_url != '':
if 'static' in self.config['url']:
self.image_url = f'{self.config["url"]["static"]}/{image_url}'
else:
- log.debug('no static url set, using main url, this could cause problems')
+ log.debug('no static url set, using main url')
self.image_url = f'{self.config["url"]["main"]}/{image_url}'
log.debug('final image url "%s"', self.image_url)
else:
@@ -139,24 +147,3 @@ class Page:
log.debug('no image url set for the page, could be because no'
' "image_url" was found in the metadata and/or no '
' "default_image" set in the config file')
-
- # if contains open graph elements
- # TODO: better handle this part
- # og_e = object graph entry
- og_elements: list[str] = list(self.__get_meta('og', []))
- if og_elements:
- log.debug('parsing og metadata')
- for og_e in og_elements:
- kv: list[str] = og_e.split(',', 1)
- if len(kv) != 2:
- log.error('invalid og syntax for "%s", needs to be "k, v"', og_e)
- sys.exit(1)
-
- k: str = kv[0].strip()
- v: str = kv[1].strip()
-
- log.debug('og element: ("%s", "%s")', k, v)
- self.og[k] = v
-
- else:
- log.debug('no tags to parse')
diff --git a/src/pyssg/plt/default.yaml b/src/pyssg/plt/default.yaml
index 0b722a6..ca2f7ad 100644
--- a/src/pyssg/plt/default.yaml
+++ b/src/pyssg/plt/default.yaml
@@ -10,12 +10,8 @@ path:
db: !join [*root, ".files"]
url:
main: "https://example.com"
- static: "https://static.example.com"
- default_image: "images/default.png"
fmt:
date: "%a, %b %d, %Y @ %H:%M %Z"
- list_date: "%b %d"
- list_sep_date: "%B %Y"
dirs:
/:
cfg:
@@ -24,5 +20,4 @@ dirs:
index: False
rss: False
sitemap: False
- exclude_dirs: []
...
\ No newline at end of file
--
cgit v1.2.3-70-g09d2