#!/usr/bin/env python
import sys
import re
import os
fileName = "output.txt"
outFile = "broken_links.html"
whitelistFile = "link-whitelist.txt"
if len(sys.argv) < 2:
print ("Enter path of input directory")
sys.exit()
scriptPath = sys.argv[0]
outputPath = sys.argv[1]
fileNamePath = outputPath + "/" + fileName
outFilePath = outputPath + "/" + outFile
whitelistFilePath = os.path.dirname(scriptPath) + "/" + whitelistFile
with open (whitelistFilePath) as w:
whLines = w.readlines()
whitelist = []
for line in whLines:
link = line.rstrip()
whitelist.append(link)
with open (fileNamePath) as f:
lines = f.readlines()
numBrokenLinks = 0
numWhiteListMatches = 0
newLines = [""]
whiteListLines = []
for line in lines:
if "[broken]" in line:
strings = line.split(" ")
link = strings[2][:-1]
link = link.strip()
if link in whitelist:
whiteListLines.append("" + strings[0] + "\n[whitelist] " + link + "
\n")
numWhiteListMatches += 1
else:
newLines.append("" + strings[0] + "\n[broken] " + link + "
\n")
numBrokenLinks += 1
newLines.insert(0,"" + str(numBrokenLinks + numWhiteListMatches) + " broken links found in Sphinx link check
\n")
newLines.insert(1,"" + str(numBrokenLinks) + " unmatched broken links
\n")
newLines.append("" + str(numWhiteListMatches) + " links matched whitelist
\n")
for line in whiteListLines:
newLines.append(line)
newLines.append("")
with open (outFilePath, "w") as outF:
for line in newLines:
outF.write(line)
print("See ./" + outFilePath + " for a detailed breakdown of broken links.")
if numBrokenLinks != 0:
print (numBrokenLinks + " detected. Exiting with error code 255.")
sys.exit(-1)
else:
print ("No unexpected broken links detected.")