Merge pull request #87 from Police-Data-Accessibility-Project/parse-r…

…esponse-59

parse_response refactor

html_tag_collector/DataClassTags.py

@@ -0,0 +1,19 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class Tags:
+    index: int = None
+    url: str = ""
+    url_path: str = ""
+    html_title: str = ""
+    meta_description: str = ""
+    root_page_title: str = ""
+    http_response: int = -1
+    h1: str = ""
+    h2: str = ""
+    h3: str = ""
+    h4: str = ""
+    h5: str = ""
+    h6: str = ""
+    div_text: str = ""

html_tag_collector/collector.py

@@ -1,3 +1,17 @@
+""" The tag collector is used to collect HTML tags and other relevant data from websites that is useful for training prediction models.
+    Information being collected includes:
+        - The URL's path
+        - HTML title
+        - Meta description
+        - The root page's HTML title
+        - HTTP response code
+        - Contents of H1-H6 header tags
+        - Contents of div tags
+"""
+
+
+from dataclasses import asdict
+from collections import namedtuple
 import json
 import ssl
 import urllib3
@@ -20,6 +34,8 @@
 
 from RootURLCache import RootURLCache
 from common import get_user_agent
+from DataClassTags import Tags
+
 
 # Define the list of header tags we want to extract
 header_tags = ["h1", "h2", "h3", "h4", "h5", "h6"]
@@ -162,27 +178,46 @@ async def get_response(session, url, index):
         except (KeyError, AttributeError):
             pass
 
-        # If the response size is greater than 10 MB
-        # or the response is an unreadable content type
-        # or the response code from the website is not in the 200s
-        if (
-            response is not None and len(response.content) > 10000000
-            or content_type is not None and any(
-                filtered_type in content_type
-                for filtered_type in ["pdf", "excel", "msword", "image", "rtf", "zip", "octet", "csv", "json"]
-            )
-            or response is not None and not response.ok
-        ):
-            # Discard the response content to prevent out of memory errors
-            if DEBUG:
-                print("Large or unreadable content discarded:", len(response.content), url)
-            new_response = requests.Response()
-            new_response.status_code = response.status_code
-            response = new_response
+        response = response_valid(response, content_type, url)
 
         return {"index": index, "response": response}
 
 
+def response_valid(response, content_type, url):
+    """Checks the response to see if content is too large, unreadable, or invalid response code. The response is discarded if it is invalid.
+
+    Args:
+        response (Response): Response object to check.
+        content_type (str): The content type returned by the website.
+        url (str): URL that was requested.
+
+    Returns:
+        Response: The response object is returned either unmodified or discarded.
+    """    
+    # If the response size is greater than 10 MB
+    # or the response is an unreadable content type
+    # or the response code from the website is not in the 200s
+    if (
+        response is not None
+        and len(response.content) > 10000000
+        or content_type is not None
+        and any(
+            filtered_type in content_type
+            for filtered_type in ["pdf", "excel", "msword", "image", "rtf", "zip", "octet", "csv", "json"]
+        )
+        or response is not None
+        and not response.ok
+    ):
+        # Discard the response content to prevent out of memory errors
+        if DEBUG:
+            print("Large or unreadable content discarded:", len(response.content), url)
+        new_response = requests.Response()
+        new_response.status_code = response.status_code
+        response = new_response
+
+    return response
+
+
 async def render_js(urls_responses):
     """Renders JavaScript from a list of urls.
 
@@ -231,70 +266,182 @@ def parse_response(url_response):
     """Parses relevant HTML tags from a Response object into a dictionary.
 
     Args:
-        url_response (list[dict]): List of dictionaries containing urls and theeir responses.
+        url_response (list[dict]): List of dictionaries containing urls and their responses.
 
     Returns:
-        list[dict]: List of dictionaries containing urls and relevant HTML tags.
+        dict: Dictionary containing the url and relevant HTML tags.
     """
-    remove_excess_whitespace = lambda s: " ".join(s.split()).strip()
-
-    tags = {}
+    tags = Tags()
     res = url_response["response"]
-    tags["index"] = url_response["index"]
+    tags.index = url_response["index"]
 
-    # Drop hostname from urls to reduce training bias
+    tags.url, tags.url_path = get_url(url_response)
+
+    tags.root_page_title = remove_excess_whitespace(root_url_cache.get_title(tags.url))
+
+    verified, tags.http_response = verify_response(res)
+    if verified is False:
+        return asdict(tags)
+
+    parser = get_parser(res)
+    if parser is False:
+        return asdict(tags)
+
+    try:
+        soup = BeautifulSoup(res.html.html, parser)
+    except (bs4.builder.ParserRejectedMarkup, AssertionError, AttributeError):
+        return asdict(tags)
+
+    tags.html_title = get_html_title(soup)
+
+    tags.meta_description = get_meta_description(soup)
+
+    tags = get_header_tags(tags, soup)
+
+    tags.div_text = get_div_text(soup)
+
+    # Prevents most bs4 memory leaks
+    if soup.html:
+        soup.html.decompose()
+
+    return asdict(tags)
+
+
+def get_url(url_response):
+    """Returns the url and url_path.
+
+    Args:
+        url_response (list[dict]): List of dictionaries containing urls and their responses.
+
+    Returns:
+        (str, str): Tuple with the url and url_path.
+    """
     url = url_response["url"][0]
-    tags["url"] = url
-    if not url.startswith("http"):
-        url = "https://" + url
-    tags["url_path"] = urlparse(url).path[1:]
+    new_url = url
+    if not new_url.startswith("http"):
+        new_url = "https://" + new_url
 
-    tags["html_title"] = ""
-    tags["meta_description"] = ""
-    tags["root_page_title"] = remove_excess_whitespace(root_url_cache.get_title(tags["url"]))
+    # Drop hostname from urls to reduce training bias
+    url_path = urlparse(new_url).path[1:]
+    # Remove trailing backslash
+    if url_path and url_path[-1] == "/":
+        url_path = url_path[:-1]
 
+    return url, url_path
+
+
+def verify_response(res):
+    """Verifies the webpage response is readable and ok.
+
+    Args:
+        res (HTMLResponse|Response): Response object to verify.
+
+    Returns:
+        VerifiedResponse(bool, int): A named tuple containing False if verification fails, True otherwise and the http response code.
+    """
+    VerifiedResponse = namedtuple("VerifiedResponse", "verified http_response")
+    # The response is None if there was an error during connection, meaning there is no content to read
     if res is None:
-        tags["http_response"] = -1
-        return tags
+        return VerifiedResponse(False, -1)
 
-    tags["http_response"] = res.status_code
+    # If the connection did not return a 200 code, we can assume there is no relevant content to read
+    http_response = res.status_code
     if not res.ok:
-        return tags
+        return VerifiedResponse(False, http_response)
+
+    return VerifiedResponse(True, http_response)
+
 
+def get_parser(res):
+    """Retrieves the parser type to use with BeautifulSoup.
+
+    Args:
+        res (HTMLResponse|Response): Response object to read the content-type from.
+
+    Returns:
+        str|bool: A string of the parser to use, or False if not readable.
+    """
+    # Attempt to read the content-type, set the parser accordingly to avoid warning messages
     try:
         content_type = res.headers["content-type"]
     except KeyError:
-        return tags
+        return False
 
+    # If content type does not contain "html" or "xml" then we can assume that the content is unreadable
     if "html" in content_type:
         parser = "lxml"
     elif "xml" in content_type:
         parser = "lxml-xml"
     else:
-        return tags
+        return False
 
-    try:
-        soup = BeautifulSoup(res.html.html, parser)
-    except (bs4.builder.ParserRejectedMarkup, AssertionError, AttributeError):
-        return tags
+    return parser
+
+
+def get_html_title(soup):
+    """Retrieves the HTML title from a BeautifulSoup object.
+
+    Args:
+        soup (BeautifulSoup): BeautifulSoup object to pull the HTML title from.
+
+    Returns:
+        str: The HTML title.
+    """
+    html_title = ""
 
     if soup.title is not None and soup.title.string is not None:
-        tags["html_title"] = remove_excess_whitespace(soup.title.string)
-    else:
-        tags["html_title"] = ""
+        html_title = remove_excess_whitespace(soup.title.string)
+
+    return html_title
 
+
+def get_meta_description(soup):
+    """Retrieves the meta description from a BeautifulSoup object.
+
+    Args:
+        soup (BeautifulSoup): BeautifulSoup object to pull the meta description from.
+
+    Returns:
+        str: The meta description.
+    """
     meta_tag = soup.find("meta", attrs={"name": "description"})
     try:
-        tags["meta_description"] = remove_excess_whitespace(meta_tag["content"]) if meta_tag is not None else ""
+        meta_description = remove_excess_whitespace(meta_tag["content"]) if meta_tag is not None else ""
     except KeyError:
-        tags["meta_description"] = ""
+        return ""
+
+    return meta_description
+
 
+def get_header_tags(tags, soup):
+    """Updates the Tags DataClass with the header tags.
+
+    Args:
+        tags (Tags): DataClass for relevant HTML tags.
+        soup (BeautifulSoup): BeautifulSoup object to pull the header tags from.
+
+    Returns:
+        Tags: DataClass with updated header tags.
+    """
     for header_tag in header_tags:
         headers = soup.find_all(header_tag)
-        # Retreives and drops headers containing links to reduce training bias
+        # Retrieves and drops headers containing links to reduce training bias
         header_content = [header.get_text(" ", strip=True) for header in headers if not header.a]
-        tags[header_tag] = json.dumps(header_content, ensure_ascii=False)
+        tag_content = json.dumps(header_content, ensure_ascii=False)
+        setattr(tags, header_tag, tag_content)
+
+    return tags
+
+
+def get_div_text(soup):
+    """Retrieves the div text from a BeautifulSoup object.
 
+    Args:
+        soup (BeautifulSoup): BeautifulSoup object to pull the div text from.
+
+    Returns:
+        str: The div text.
+    """
     # Extract max 500 words of text from HTML <div>'s
     div_text = ""
     MAX_WORDS = 500
@@ -307,14 +454,22 @@ def parse_response(url_response):
             else:
                 break  # Stop adding text if word limit is reached
 
-    # truncate to 5000 characters in case of run-on 'words'
-    tags["div_text"] = div_text[:MAX_WORDS * 10]
+    # Truncate to 5000 characters in case of run-on 'words'
+    div_text = div_text[: MAX_WORDS * 10]
 
-    # Prevents most bs4 memory leaks
-    if soup.html:
-        soup.html.decompose()
+    return div_text
 
-    return tags
+
+def remove_excess_whitespace(s):
+    """Removes leading, trailing, and excess adjacent whitespace.
+
+    Args:
+        s (str): String to remove whitespace from.
+
+    Returns:
+        str: Clean string with excess whitespace stripped.
+    """
+    return " ".join(s.split()).strip()
 
 
 def collector_main(df, render_javascript=False):