# HG changeset patch
# User Laman
# Date 2018-12-15 09:20:11
# Node ID a1e2cc709d537226fc23c32b7a46510e22f6e281
# Parent  9df0d1a019c5c96cf56425f4f19a9844a2e50772

photo collector

diff --git a/exp/collector.py b/exp/collector.py
new file mode 100644
--- /dev/null
+++ b/exp/collector.py
@@ -0,0 +1,33 @@
+import os
+import sys
+import random
+import time
+import urllib.request
+
+from lxml import html
+
+
+targetDir=""
+links=sys.argv
+for link in links:
+	time.sleep(1+random.random())
+	with urllib.request.urlopen(link) as req:
+		if req.getcode()!=200:
+			print("error when downloading {0}".format(link),file=sys.stderr)
+			continue
+		doc_str=req.read().decode("utf8")
+
+	doc=html.fromstring(doc_str)
+
+	license=doc.xpath("//div[@class='photo-license-info']//span/text()")[0].strip()
+	imgUrl=doc.xpath("//meta[@property='og:image']/@content")[0]
+	filename=imgUrl.split("/")[-1]
+
+	with urllib.request.urlopen(imgUrl) as req:
+		if req.getcode()!=200:
+			print("error when downloading {0}".format(link),file=sys.stderr)
+			continue
+		with open(os.path.join(targetDir,filename),mode="wb") as f:
+			f.write(req.read())
+
+	print("\t".join(map(str, [link,filename,license])))