1 files changed, 40 insertions, 0 deletions
diff --git a/python/parse_preqin.py b/python/parse_preqin.py
new file mode 100644
index 00000000..6dfcd0a0
--- /dev/null
+++ b/python/parse_preqin.py
@@ -0,0 +1,40 @@
+import os
+import bs4
+import csv
+import re
+
+root = "/home/share/serenitas/Fund Raising/Preqin/Investors/"
+filelist = [f for f in os.listdir(root) if f.endswith('htm')]
+readdress=re.compile("address")
+retel=re.compile("tel")
+refax=re.compile("fax")
+recat=re.compile("fund_category")
+reabout=re.compile("about")
+fh2 = open(os.path.join(root, "investors.csv"), "w")
+csvwriter = csv.writer(fh2)
+headers = ['filename','name', 'address', 'tel', 'fax','category','about']
+csvwriter.writerow(headers)
+d = {}
+sentinel={}
+for f in filelist:
+    with open(os.path.join(root, f)) as fh:
+        soup = bs4.BeautifulSoup(fh)
+        d['filename']=f
+        d['name']=soup.findAll("h1")[1].text.strip()
+        if d['name'] in sentinel:
+            continue
+        else:
+            sentinel[d['name']]=1
+        d['address']=soup.find(id=readdress).text
+        try:
+            d['tel']=soup.find(id=retel).text
+        except AttributeError:
+            d['tel']=''
+        d['about']=soup.find(id=reabout).text.replace("\n","").strip()
+        try:
+            d['fax']=soup.find(id=refax).text
+        except AttributeError:
+            d['fax']=''
+        d['category']=soup.find(id=recat).text
+        csvwriter.writerow([d[k] for k in headers])
+fh2.close()