summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorsstvinc2 <sstvinc2@gmail.com>2017-02-15 16:14:38 -0600
committersstvinc2 <sstvinc2@gmail.com>2017-02-15 16:14:38 -0600
commit38483987b2389b92ca06ac1b409f358ecd4fa991 (patch)
tree072ae9e6d913b0fe69c140d47cbd040135cab91e
parent8e87842bdbd8525c4fa6ec8f1bd95aa42ab9318b (diff)
Changed randomization algorith for H2 and H3; fully implemented H3
-rw-r--r--html_template/template.html32
-rw-r--r--main.py17
-rw-r--r--parser.py1
-rw-r--r--unbiasedFunctions.py96
4 files changed, 55 insertions, 91 deletions
diff --git a/html_template/template.html b/html_template/template.html
index befaaff..1c2e858 100644
--- a/html_template/template.html
+++ b/html_template/template.html
@@ -109,35 +109,51 @@
<div id="bottom-stories">
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-1.html">xxTitle3-1xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-1xx', '_blank')">xxTitle3-1xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-2.html">xxTitle3-2xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-2xx', '_blank')">xxTitle3-2xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-3.html">xxTitle3-3xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-3xx', '_blank')">xxTitle3-3xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-4.html">xxTitle3-4xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-4xx', '_blank')">xxTitle3-4xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-5.html">xxTitle3-5xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-5xx', '_blank')">xxTitle3-5xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-6.html">xxTitle3-6xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-6xx', '_blank')">xxTitle3-6xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-7.html">xxTitle3-7xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-7xx', '_blank')">xxTitle3-7xx</a>
</div>
<div class="bottom-story">
- <a target="_blank" href="redirects/h3-8.html">xxTitle3-8xx</a>
+ <a target="_blank" href="" onclick="window.open('xxURL3-8xx', '_blank')">xxTitle3-8xx</a>
+ </div>
+
+ <div class="bottom-story">
+ <a target="_blank" href="" onclick="window.open('xxURL3-9xx', '_blank')">xxTitle3-9xx</a>
+ </div>
+
+ <div class="bottom-story">
+ <a target="_blank" href="" onclick="window.open('xxURL3-10xx', '_blank')">xxTitle3-10xx</a>
+ </div>
+
+ <div class="bottom-story">
+ <a target="_blank" href="" onclick="window.open('xxURL3-11xx', '_blank')">xxTitle3-11xx</a>
+ </div>
+
+ <div class="bottom-story">
+ <a target="_blank" href="" onclick="window.open('xxURL3-12xx', '_blank')">xxTitle3-12xx</a>
</div>
</div>
diff --git a/main.py b/main.py
index cf68d01..92f96ae 100644
--- a/main.py
+++ b/main.py
@@ -27,23 +27,11 @@ def run():
cbs=buildCBS()
sourceList.append(cbs)
-
- '''
- sourceList.append(NewsSource('The Blaze',
- 'http://theblaze.com',
- ['<a class="gallery-link" href="'],
- ['</figure>\n\n<figure class="gallery-item">', 'href="'],
- [],
- '<!-- home -->', '<!-- loop-home -->',
- '<!-- home -->', '<!-- loop-home -->',
- None, None))
- '''
-
wkl=buildWeeklyStandard()
sourceList.append(wkl)
- nyt=buildNYT()
- sourceList.append(nyt)
+ #nyt=buildNYT()
+ #sourceList.append(nyt)
fox=buildFoxNews()
sourceList.append(fox)
@@ -53,6 +41,7 @@ def run():
#build the output file HTML
outputHTML=buildOutput(newsSourceArr)
+
#print the output file HTML
printOutputHTML(outputHTML, '/var/www/html/index.html')
diff --git a/parser.py b/parser.py
index b9a05b9..53b3261 100644
--- a/parser.py
+++ b/parser.py
@@ -178,7 +178,6 @@ def blazeFixDesc(articleArr):
desc=TAG_RE.sub('', desc)
desc=desc.replace('\n', ' ')
desc=desc[:144]
- print(desc+'\n\n')
articleArr[i].description=desc
return articleArr
diff --git a/unbiasedFunctions.py b/unbiasedFunctions.py
index 733d6ba..46723cd 100644
--- a/unbiasedFunctions.py
+++ b/unbiasedFunctions.py
@@ -54,58 +54,6 @@ def buildArticle(url, sourceName):#, titleDelStart, titleDelEnd, imgDelStart, im
return None
-#take in a read main source file (e.g. from nytimes.com) and return lists of the urls for stories
-def extractURLs(content, source):
- h1s=[]
- h2s=[]
- h3s=[]
-
- try:
- h1=content
- if source.h1SectionDividerStart!=None:
- h1=h1.split(source.h1SectionDividerStart)[1]
- if source.h1SectionDividerEnd!=None:
- h1=h1.split(source.h1SectionDividerEnd)[0]
- for delim in source.h1DelStart:
- h1=h1.split(delim)[1]
- h1=h1.split(source.h1DelEnd)[0]
- if '.com' not in h1:
- if source.stubURL!=None:
- h1=source.stubURL+h1
- else:
- h1=source.url+h1
- h1s.append(h1)
- except:
- print("Parse error in extractURLs: "+source.name+" h1")
- h1s=None
-
- try:
- h2=content
- if source.h2SectionDividerStart!=None:
- h2=h2.split(source.h2SectionDividerStart, 1)[1]
- if source.h2SectionDividerEnd!=None:
- h2=h2.split(source.h2SectionDividerEnd, 1)[0]
-
- while source.h2DelStart[0] in h2:
- x=h2
- for delim in source.h2DelStart:
- x=x.split(delim)[1]
- h2=h2.split(delim, 1)[1]
- x=x.split(source.h2DelEnd)[0]
- h2=h2.split(source.h2DelEnd, 1)[1]
- if '.com' not in x:
- if source.stubURL!=None:
- x=source.stubURL+x
- else:
- x=source.url+x
- h2s.append(x)
- except:
- print("Parse error in extractURLs: "+source.name+" h2")
- h2s=None
-
- return h1s, h2s, h3s
-
-
def buildOutput(newsSourceArr):
#read in the template html file
f=open('html_template/template.html', 'r')
@@ -114,7 +62,23 @@ def buildOutput(newsSourceArr):
#set the random order for sources
h1RandomSources=random.sample(range(len(newsSourceArr)), 4)
- h2RandomSources=random.sample(range(len(newsSourceArr)), 6)
+ #For h2s and h3s, select N random sources (can repeat), then
+ #a non-repetitive random article from within that source
+ h2RandomPairs=[]
+ while len(h2RandomPairs) < 6:
+ x=random.sample(range(len(newsSourceArr)), 1)[0]
+ y=random.sample(range(len(newsSourceArr[x].h2Arr)), 1)[0]
+ pair=[x,y]
+ if not pair in h2RandomPairs:
+ h2RandomPairs.append(pair)
+ h3RandomPairs=[]
+ while len(h3RandomPairs) < 12:
+ x=random.sample(range(len(newsSourceArr)), 1)[0]
+ y=random.sample(range(len(newsSourceArr[x].h3Arr)), 1)[0]
+ pair=[x,y]
+ if not pair in h3RandomPairs:
+ h3RandomPairs.append(pair)
+
#replace html template locations with data from newsSourceArr
for i in range(len(h1RandomSources)):
@@ -122,11 +86,6 @@ def buildOutput(newsSourceArr):
randomArticle=random.sample(range(len(source.h1Arr)), 1)[0]
article=source.h1Arr[randomArticle]
template=template.replace('xxURL1-'+str(i+1)+'xx', article.url)
- '''
- r=open('/var/www/html/redirects/h1-'+str(i+1)+'.html', 'w')
- r.write('<html><head><script type="text/javascript">window.location="'+article.url+'"</script></head></html>')
- r.close()
- '''
template=template.replace('xxTitle1-'+str(i+1)+'xx', article.title)
template=template.replace('xxImg1-'+str(i+1)+'xx', article.img)
desc=article.description
@@ -136,20 +95,21 @@ def buildOutput(newsSourceArr):
desc=' '.join(desc)+' ...'
template=template.replace('xxDesc1-'+str(i+1)+'xx', desc)
-
- for i in range(len(h2RandomSources)):
- source=newsSourceArr[h2RandomSources[i]]
- randomArticle=random.sample(range(len(source.h2Arr)), 1)[0]
- article=source.h2Arr[randomArticle]
+ for i in range(len(h2RandomPairs)):
+ pair=h2RandomPairs[i]
+ article=newsSourceArr[pair[0]].h2Arr[pair[1]]
template=template.replace('xxURL2-'+str(i+1)+'xx', article.url)
- '''
- r=open('/var/www/html/redirects/h2-'+str(i+1)+'.html', 'w')
- r.write('<html><head><script type="text/javascript">window.location="'+article.url+'"</script></head></html>')
- r.close()
- '''
template=template.replace('xxTitle2-'+str(i+1)+'xx', article.title)
template=template.replace('xxImg2-'+str(i+1)+'xx', article.img)
+ for i in range(len(h3RandomPairs)):
+ pair=h3RandomPairs[i]
+ article=newsSourceArr[pair[0]].h3Arr[pair[1]]
+ template=template.replace('xxURL3-'+str(i+1)+'xx', article.url)
+ template=template.replace('xxTitle3-'+str(i+1)+'xx', article.title)
+ template=template.replace('xxImg3-'+str(i+1)+'xx', article.img)
+
+
sourcesStr=''
for i in range(len(newsSourceArr)-1):
sourcesStr+=newsSourceArr[i].name+', '