diff options
| author | grothedev <grothedev@gmail.com> | 2022-04-02 15:22:11 -0500 |
|---|---|---|
| committer | grothedev <grothedev@gmail.com> | 2022-04-02 15:22:11 -0500 |
| commit | 7517de1aacdcba774be800f5d8d95934da4759aa (patch) | |
| tree | ed12edac8e156c7513f00b4e34e458e6a4be3a95 | |
| parent | bd350248819996d31e43418ff881faeb3c223835 (diff) | |
updates to 4chan query script
| -rwxr-xr-x | 4chan_search/query.py | 23 |
1 files changed, 20 insertions, 3 deletions
diff --git a/4chan_search/query.py b/4chan_search/query.py index f3fc5ce..414bae8 100755 --- a/4chan_search/query.py +++ b/4chan_search/query.py @@ -30,20 +30,32 @@ def processCatalog(catalog, b): results_content.append(catalog[i]['threads'][j]['last_replies'][k]['com']) #imgs were already retrieved from OP grab +def repliesSort(catalog): + result = [] + for i in range(0, len(catalog)): + for j in range(0, len(catalog[i]['threads'])): + url = "https://boards.4channel.org/"+b+"/thread/"+str(catalog[i]['threads'][j]['no']) + result.append((url, catalog[i]['threads'][j]['replies'])) + + result.sort(key=lambda v: v[1]) + return result + bods = ['a', 'c', 'w', 'm', 'cgl', 'cm', 'f', 'n', 'jp', 'vp', 'v', 'vg', 'vr', 'co', 'g', 'tv', 'k', 'o', 'an', 'tg', 'sp', 'asp', 'sci', 'int', 'out', 'toy', 'biz', 'i', 'po', 'p', 'ck', 'ic', 'wg', 'mu', 'fa', '3', 'gd', 'diy', 'wsg', 's', 'trv', 'fit', 'x', 'lit', 'adv', 'lgbt', 'mlp', 'b', 'r', 'r9k', 'pol', 'soc', 's4s'] abods = ['hc', 'hm', 'h', 'e', 'u', 'd', 'y', 't', 'hr', 'gif'] ##################################################################### +r = True #just for debugging right now -if len(sys.argv) < 2 or sys.argv[1][0] == '-': +if not r and (len(sys.argv) < 2 or sys.argv[1][0] == '-'): print('This program will give you the URLs of all 4chan posts that contain the given search word, either on the entire site or on a select board. You can also use wildcard as search word for all posts of board/site') print('you must provide a search word, or \"*\" for any word.') print('Usage: ./query.py <searchword> [board]') sys.exit(0) -wod = sys.argv[1] #the search keyword +if not r: + wod = sys.argv[1] #the search keyword bod = '' if len(sys.argv) > 2: bod = sys.argv[2] #the board if interest, if given @@ -54,12 +66,17 @@ results_img = [] #URLs of all images containing the keyword if bod == '': print('searching all boards') + repl_res = [] for b in bods: res = requests.get("https://a.4cdn.org/"+b+"/catalog.json") - processCatalog(json.loads(res.text), b) + repl_res.append(repliesSort(json.loads(res.text))) + #processCatalog(json.loads(res.text), b) + repl_res.sort(key=lambda v: v[1]) + print(repl_res) else: print('searching board ' + bod) res = requests.get("https://a.4cdn.org/"+bod+"/catalog.json") + print(repliesSort(json.loads(res.text))) processCatalog(json.loads(res.text), bod) |
