I'm not the original @RepostStatistics, but I'm their spiritual successor. The original bot hasn't been active for a few years, but they were kind enough to leave the source code for the project on GitHub. I thought the bot was genuinely useful, so I decided to update it for Python 3, fix what no longer works, and start indexing reposts again. Original source code: https://github.com/SleepProgger/RepostStatistics Fork: https://github.com/irnutsmurt/RepostStatisticsRedux tl;dr: bot who checks for reposts Comments containing repost statistics, such as "Image last seen..." and title-similarity stars, are automatically posted by the bot. ########################### WTF is this? ########################### This bot tries to detect when an image has been posted before by generating hashes that describe the image and comparing them against previously indexed images. When a match is found and certain conditions are met, the bot can comment with information about when that image was previously seen. The stars represent how similar the titles of the two posts are using Levenshtein distance, essentially how many characters would need to be changed, inserted, or removed to turn one title into the other. Examples below: ☆☆☆☆☆ = little or no title similarity ★★★★★ = nearly or exactly the same title If you don't want this bot commenting on your posts, comment @RepostStatisticsRedux blockme and you'll automatically be added to the blocklist. ########################### For programmers ########################### The bot is written in Python 3. Still images are compared using perceptual image hashes, including dHash and aHash at 256 bits, to reduce false positives. Animated images are handled separately. This version is based on the original RepostStatistics source code that was written in python2 but has been updated and expanded upon where necessary for Python3, current libraries, and the current Imgur API.
222
66
4620