def read(self, name: str, cache: DirectoryCache):
key = f"{name}_data.txt"
content = cache.read(key)
if content == None:
self.df = dataframe_from_text(content)
key = f"{name}_source.{self.content_type}"
self.content = cache.read(key)
class PageCompare():
def __init__(self, work_dir_a: str, work_dir_b: str, out_dir: str):
self.cache_a = DirectoryCache(work_dir_a)
self.cache_b = DirectoryCache(work_dir_b)
self.out_dir = out_dir
def process_all(self):
ignore_list = ["main_sheet", "WV"]
if not os.path.exists(self.out_dir): os.makedirs(self.out_dir)
for fn in self.cache_a.list_html_files():
x = fn.replace(".html", "")
if x in ignore_list: continue
content_a = self.cache_a.read(fn)
content_b = self.cache_b.read(fn)
if content_a == content_b:
logger.info(f"=================| {fn}")
logger.info(" data is SAME")
continue
fn_a = os.path.join(self.out_dir, f"{x}_A.html")
fn_b = os.path.join(self.out_dir, f"{x}_B.html")
if os.path.exists(fn_a): os.remove(fn_a)
if os.path.exists(fn_b): os.remove(fn_b)
if content_a == content_b:
logger.info(f"=================| {fn}")
logger.info(" data is FIXED")
continue
doc_a = html.fromstring(content_a)
doc_b = html.fromstring(content_b)
remove_identical_nodes(doc_a, doc_b)
str_a = html.tostring(doc_a, pretty_print=True)
str_b = html.tostring(doc_b, pretty_print=True)
logger.info(f"=================| {fn}")
logger.info(" data is different")
with open(fn_a, "wb") as f:
f.write(str_a)
with open(fn_b, "wb") as f:
f.write(str_b)
def read(self, cache: DirectoryCache, name: str):
if len(self.names) == 0: raise Exception("No sources")
content = cache.read(name)
if content == None: return
df = dataframe_from_text(content)
df.index = df.names
missing = ~df.index.isin(self.names)
if len(missing) > 0:
df.loc[missing, "status"] = "removed"
df.reset_index(inplace=True, drop=True)
self.df_status = df
def write(self, name: str, cache: DirectoryCache, change_list: ChangeList):
key = f"{name}_data.txt"
new_content = dataframe_to_text(self.df)
old_content = cache.read(key)
if old_content != new_content:
cache.write(key, new_content)
key = f"{name}_source.{self.content_type}"
cache.write(key, self.content)
change_list.record_changed(name, "source", self.endpoint)
else:
change_list.record_unchanged(name, "source", self.endpoint)
class DataPipeline():
def __init__(self, config: DataPipelineConfig):
self.config = config
self.change_list: ChangeList = None
base_dir = config.base_dir
self.cache_sources = DirectoryCache(os.path.join(base_dir, "sources"))
self.cache_raw = DirectoryCache(os.path.join(base_dir, "raw"))
self.cache_clean = DirectoryCache(os.path.join(base_dir, "clean"))
self.cache_extract = DirectoryCache(os.path.join(base_dir, "extract"))
self.cache_diff = DirectoryCache(os.path.join(base_dir, "diff"))
self.url_manager = UrlManager(config.headless, config.browser)
self.sources: UrlSources = None
self._capture: SpecializedCapture = None
def get_capture(self) -> SpecializedCapture:
if self._capture == None:
publish_dir = os.path.join(self.config.base_dir, 'captive-browser')
driver = self.url_manager._captive.driver if self.url_manager._captive else None
self._capture = SpecializedCapture(self.config.temp_dir,
publish_dir, driver)
return self._capture
def shutdown_capture(self):
if self._capture != None:
self._capture.close()
if self.config.auto_push:
self._capture.publish()
self._capture = None
def update_sources(self):
" update the remote url sources "
manager = UrlSourceManager(self.cache_sources)
self.sources = manager.update_sources("scan")
def process(self) -> Dict[str, str]:
" run the pipeline "
self.url_manager.reset()
self.change_list = ChangeList(self.cache_raw)
host = get_host()
print(
f"=== run started on {host} at {udatetime.to_logformat(self.change_list.start_date)}"
)
self.change_list.start_run()
try:
if self.sources == None:
raise Exception("Sources not provided")
src = self.sources.items[0]
if src.name != "google-states-csv":
raise Exception(
f"Expected first source to be google-states-csv, not {src.name}"
)
return self._main_loop(src, self.change_list)
except Exception as ex:
logger.exception(ex)
self.change_list.abort_run(ex)
finally:
self.change_list.finish_run()
self.shutdown_capture()
logger.info(
f" [in-memory content cache took {self.url_manager.size*1e-6:.1f} MBs"
)
logger.info(
f"run finished on {host} at {udatetime.to_logformat(self.change_list.start_date)}"
)
def format_html(self, rerun=False):
" format raw html "
is_first = False
for key in self.cache_raw.list_html_files():
if key == "index.html": continue
if key == "google_sheet.html": continue
if rerun or not self.cache_raw.exists(key):
if is_first:
logger.info(f"format existing files...")
is_first = False
logger.info(f" format {key}")
local_raw_content = self.cache_raw.read(key)
formater = HtmlFormater()
local_clean_content = formater.format(None, local_raw_content)
self.cache_raw.write(key, local_clean_content)
def clean_html(self, rerun=False):
" generate clean files from existing raw html "
is_first = False
for key in self.cache_raw.list_html_files():
if key == "index.html": continue
if key == "google_sheet.html": continue
if rerun or not self.cache_clean.exists(key):
if is_first:
logger.info(f"clean existing files...")
is_first = False
logger.info(f" clean {key}")
local_raw_content = self.cache_raw.read(key)
cleaner = HtmlCleaner()
local_clean_content = cleaner.clean(local_raw_content)
self.cache_clean.write(key, local_clean_content)
def extract_html(self, rerun=False):
" generate extract files from existing clean html "
self.change_list = ChangeList(self.cache_raw)
self.change_list.load()
is_first = False
for key in self.cache_clean.list_html_files():
if key == "index.html": continue
if key == "google_sheet.html": continue
if rerun or not self.cache_extract.exists(key):
if is_first:
logger.info(f"extract existing files...")
is_first = False
logger.info(f" extract {key}")
local_clean_content = self.cache_clean.read(key)
item = self.change_list.get_item(key)
if item == None:
logger.warning(" skip because it is a new item")
continue
extracter = HtmlExtracter()
local_extract_content = extracter.extract(
local_clean_content, item)
self.cache_extract.write(key, local_extract_content)
def _main_loop(self, source: UrlSource,
change_list: ChangeList) -> Dict[str, str]:
def remove_duplicate_if_exists(location: str, source: str,
other_state: str):
key = location + ".html"
self.cache_raw.remove(key)
self.cache_clean.remove(key)
change_list.record_duplicate(key, source,
f"duplicate of {other_state}")
if self.config.capture_image:
c = self.get_capture()
c.remove(location)
def fetch_if_changed(location: str,
source: str,
xurl: str,
skip: bool = False) -> bool:
key = location + ".html"
if xurl == "" or xurl == None or xurl == "None":
change_list.record_skip(key, source, xurl, "missing url")
return
mins = change_list.get_minutes_since_last_check(key)
if self.config.trace:
logger.info(f" checked {key} {mins:.1f} minutes ago")
if mins < 15.0:
if self.config.rerun_now:
logger.info(f"{key}: checked {mins:.1f} mins ago")
else:
logger.info(
f"{key}: checked {mins:.1f} mins ago -> skip b/c < 15 mins"
)
change_list.temporary_skip(key, source, xurl,
"age < 15 mins")
return False
if skip:
change_list.record_skip(key, source, xurl, "skip flag set")
return False
if self.config.trace: logger.info(f"fetch {xurl}")
remote_raw_content, status = self.url_manager.fetch(xurl)
is_bad, msg = is_bad_content(remote_raw_content)
if is_bad:
change_list.record_failed(key, source, xurl, msg)
return False
if status > 300:
change_list.record_failed(location, source, xurl,
f"HTTP status {status}")
return False
remote_raw_content = remote_raw_content.replace(b"\r", b"")
formater = HtmlFormater()
remote_raw_content = formater.format(xurl, remote_raw_content)
local_clean_content = self.cache_clean.read(key)
cleaner = HtmlCleaner()
remote_clean_content = cleaner.clean(remote_raw_content)
if local_clean_content != remote_clean_content:
self.cache_raw.write(key, remote_raw_content)
self.cache_clean.write(key, remote_clean_content)
change_list.record_changed(key, source, xurl)
item = change_list.get_item(key)
formatter = HtmlFormater()
remote_raw_content = formatter.format(xurl, remote_raw_content)
extracter = HtmlExtracter()
remote_extract_content = extracter.extract(
remote_clean_content, item)
self.cache_extract.write(key, remote_extract_content)
if self.config.capture_image:
c = self.get_capture()
c.screenshot(key, f"Screenshot for {location}", xurl)
else:
change_list.record_unchanged(key, source, xurl)
return False
# -- get urls to hit
if source.status != "valid":
raise Exception(f"URL source {source.name} status is not valid")
df_config = source.df
if df_config is None:
raise Exception(
f"URL source {source.name} does not have any data loaded")
# -- fetch pages
skip = False
err_cnt = 0
for idx, r in df_config.iterrows():
location = r["location"]
source = r["source_name"]
general_url = r["main_page"]
data_url = r["data_page"]
if general_url == None and data_url == None:
logger.warning(f" no urls for {location} -> skip")
change_list.record_skip(location)
continue
if idx % 10 == 1: change_list.save_progress()
if general_url != None:
try:
fetch_if_changed(location, source, general_url, skip=skip)
except Exception as ex:
err_cnt += 1
if err_cnt > 10: break
change_list.record_failed(location, source, general_url,
"Exception in code")
logger.exception(ex)
logger.error(" error -> continue to next page")
if data_url != None:
if general_url == data_url:
remove_duplicate_if_exists(location + "_data", source,
location)
else:
try:
fetch_if_changed(location + "_data",
source,
data_url,
skip=skip)
except Exception as ex:
err_cnt += 1
if err_cnt > 10: break
change_list.record_failed(location, source,
general_url,
"Exception in code")
logger.exception(ex)
logger.error(" error -> continue to next page")
if err_cnt > 10:
logger.error(f" abort run due to {err_cnt} errors")
change_list.write_html_to_cache(self.cache_raw, "RAW")
change_list.write_html_to_cache(self.cache_clean, "CLEAN")
change_list.write_html_to_cache(self.cache_extract, "EXTRACT")
class PageParser():
def __init__(self, work_dir: str):
self.cache = DirectoryCache(work_dir)
def strip_attributes(self, x: html.Element):
x.attrib.clear()
for y in x:
self.strip_attributes(y)
def extract_tables(self, tables: List) -> List[ContentTable]:
result = []
for t in tables:
self.strip_attributes(t)
x = ContentTable(t)
result.append(x)
return result
def write_as_text(self, foutput, name: str, tables: List[ContentTable]):
foutput.write(f"{name}\n")
for i in range(len(tables)):
foutput.write(f"{name}\tTable {i+1}\n")
t = tables[i]
for r in t.rows:
foutput.write(f"{name}\tTable {i+1}")
for c in r:
foutput.write("\t")
if c != None:
c2 = unidecode(c)
foutput.write(c2)
foutput.write(f"\n")
foutput.write(f"\n")
def write_as_html(self, foutput, name: str, url: str,
tables: List[ContentTable], html_doc: html.Element):
s = html.Element("div")
h = html.Element("h1")
h.text = name
s.append(h)
m = html.Element("div")
m.text = self.cache.read_date_time_str(name + ".html")
s.append(m)
for t in tables:
s.append(t.new_element)
x = html.Element("br")
s.append(x)
a = html.Element("a")
a.attrib["href"] = url
a.text = url
s.append(a)
h = html.Element("html")
h.append(html.Element("body"))
h[0].append(deepcopy(s))
foutput.write(html.tostring(h, pretty_print=True))
html_doc.append(s)
html_doc.append(html.Element("hr"))
def write_miss_to_html(self, name: str, url: str, msg: str,
html_doc: html.Element):
s = html.Element("div")
h = html.Element("h1")
h.text = name
s.append(h)
m = html.Element("div")
m.text = self.cache.read_date_time_str(name + ".html")
s.append(m)
m = html.Element("span")
m.text = msg
s.append(m)
x = html.Element("br")
s.append(x)
a = html.Element("a")
a.attrib["href"] = url
a.text = url
s.append(a)
html_doc.append(s)
html_doc.append(html.Element("hr"))
def process_all(self):
cl = ChangeList(self.cache)
url_dict = cl.read_urls_as_dict()
ignore_list = ["main_sheet", "WV"]
missing_list = [
"WI", "VA", "UT", "TN", "SC", "RI", "PA", "NJ", "NE", "ND",
"ND_data", "NC", "MS", "MO", "MN", "KY", "KS", "IA", "HI", "GA",
"DC_data", "AZ", "AL"
]
table_list = [
"WI_data", "WA", "VT", "TX", "SD", "SC_data", "OR", "OK", "OH",
"NY", "NV", "NM", "NJ_data", "NH", "NC_data", "MT", "MN_data",
"MI", "ME", "MD", "MA_data", "MA", "CO", "LA", "IN", "IL", "ID",
"FL", "DE", "DC", "CT", "CO_data", "CA", "AR", "AK_data", "AK"
]
out_file = os.path.join(self.cache.work_dir, "results.txt")
old_date = self.cache.read_old_date()
foutput = open(out_file, "w")
foutput.write(f"Data Scanned at\t{old_date}\n")
foutput.write(f"STATE RESULTS\n\n")
html_out_dir = os.path.join(self.cache.work_dir, "tables")
if not os.path.exists(html_out_dir): os.makedirs(html_out_dir)
html_out_file = os.path.join(self.cache.work_dir, "tables",
f"combined.html")
html_doc = html.Element("html")
html_doc.append(html.Element("body"))
html_doc[0].append(html.Element("span"))
html_doc[0][0].text = f"data scraped at {old_date}"
html_doc[0].append(html.Element("hr"))
for fn in self.cache.list_html_files():
x = fn.replace(".html", "")
if x in ignore_list: continue
logger.info(f"=================| {fn}")
content = self.cache.read(fn)
tree = html.fromstring(content)
tables = tree.xpath('//table')
if len(tables) > 0:
if x in missing_list:
foutput.write(f"{x}\t*** Found unexpected tables\n\n")
logger.warning(f" found {len(tables)} unexpected tables")
xlist = self.extract_tables(tables)
xlist2 = [x for x in xlist if x.contains_data()]
if len(xlist2) == 0:
foutput.write(f"{x}\tNo data tables\n\n")
self.write_miss_to_html(x, url_dict[x], "No data tables",
html_doc)
continue
#xlist = self.remove_nondata_tables(xlist)
self.write_as_text(foutput, x, xlist2)
html_out_file = os.path.join(self.cache.work_dir, "tables",
f"{x}.html")
with open(html_out_file, "wb") as foutput2:
self.write_as_html(foutput2, x, url_dict[x], xlist2,
html_doc)
else:
if x in table_list:
foutput.write(f"{x}\t*** Missing expected tables\n\n")
logger.warning(f" missing tables")
else:
foutput.write(f"{x}\tNo tables in data\n\n")
self.write_miss_to_html(x, url_dict[x], "No tables", html_doc)
html_out_file = os.path.join(self.cache.work_dir, "tables",
f"combined.html")
with open(html_out_file, "wb") as foutput2:
foutput2.write(html.tostring(html_doc, pretty_print=True))