parserfunctions.error_log_entry(cur, conn, mysql_log_name,
curr_time, pubshort, homepage,
seriousness, message)
#continue
##### Get Most Viewed
try:
mostviewed_1to5 = [
link.find("a", href=re.compile(pattern)) for link in
document_soup.find("li", class_="most-popular").find_all("li")
]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(
link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name,
curr_time, pubshort, homepage,
seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(
mostviewed_linklist)
except:
pop_top_5_links = []
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time,
pubshort, homepage, seriousness,
message)
#continue
##### Get Most Viewed Items
try:
mostviewed_1to5 = [
link.find("a", href=link_pattern) for link in document_soup.find(
"div", id="river-container").find_all("li")
]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(
link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time,
pubshort, homepage, seriousness,
message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(
mostviewed_linklist)
except:
pop_top_5_links = []
except:
pass
if layout is None:
message = "Failed to detect the layout"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
##### Get Most Viewed
try:
mostviewed_1to5 = [link.find("a", href=re.compile(pattern)) for link in document_soup.find("div", id="mostreademailed").find_all("li")] # ID is most e-mailed, but page states most read
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(mostviewed_linklist)
except:
pop_top_5_links = []
message = "Failed to get a POP link"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
parserfunctions.error_log_entry(cur, conn, mysql_log_name,
curr_time, pubshort, homepage,
seriousness, message)
#continue
##### Get Most Viewed
try:
mostviewed_1to5 = [
link.find("a", href=re.compile(pattern)) for link in
document_soup.find("div", id="post_most").find_all("li")
]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(
link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name,
curr_time, pubshort, homepage,
seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(
mostviewed_linklist)
except:
pop_top_5_links = []
except:
pass
if layout is None:
message = "Failed to detect the layout"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
##### Get Most Viewed
try:
mostviewed_1to5 = [link.find("a", href=re.compile(pattern)) for link in document_soup.find("div", id="post_most").find_all("li")]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(mostviewed_linklist)
except:
pop_top_5_links = []
message = "Failed to get a POP link"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
except:
pass
if layout is None:
message = "Failed to detect the layout"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
##### Get Most Viewed
try:
mostviewed_1to5 = [link.find("a", href=re.compile(pattern)) for link in document_soup.find("li", class_="most-popular").find_all("li")]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(mostviewed_linklist)
except:
pop_top_5_links = []
message = "Failed to get a POP link"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name, curr_time, pubshort, homepage, seriousness, message)
if layout is None:
print("Failed to detect the layout for document %s" % (homepage))
continue
##### Get Most Viewed
try:
mostviewed_1to5 = [
link.find("a", href=re.compile(pattern))
for link in document_soup.find("div", id="most-popular-tabs").
find("div", class_="tab").find_all("li")
]
mostviewed_linklist = []
for link in mostviewed_1to5:
try:
link = link.get("href")
mostviewed_linklist = parserfunctions.linklist_actions(
link, mostviewed_linklist)
except:
pass
except:
message = "Failed at getting MV link list"
seriousness = 2
parserfunctions.error_log_entry(cur, conn, mysql_log_name,
curr_time, pubshort, homepage,
seriousness, message)
#continue
try:
pop_top_5_links = parserfunctions.get_top_5_list_links(
mostviewed_linklist)
except:
pop_top_5_links = []