From a0ebdaf4d3bc0702a8534f1a46afa9a178099583 Mon Sep 17 00:00:00 2001 From: siddharth Date: Sun, 17 Apr 2022 02:41:16 -0400 Subject: feed: update Process Add capability to filter entries based on title. --- feed/feed.go | 33 +++++++++++++++++++++++---------- 1 file changed, 23 insertions(+), 10 deletions(-) (limited to 'feed') diff --git a/feed/feed.go b/feed/feed.go index 25c5750..a3b125c 100644 --- a/feed/feed.go +++ b/feed/feed.go @@ -1,5 +1,5 @@ // SPDX-License-Identifier: ISC -// Copyright © 2021 siddharth +// Copyright © 2022 siddharth package feed @@ -18,13 +18,14 @@ import ( ) type Feed struct { - Id string `json:"id"` - Source string `json:"source"` - Schema string `json:"schema"` - Last int `json:"last"` - YDLPath string - DumpDir string - Entries []schema.Entry + Id string + Source string + Schema string + Last int + TitleContains string `json:"title-contains"` + YDLPath string + DumpDir string + Entries []schema.Entry } func (feed *Feed) Validate(baseDumpDir string) error { @@ -124,10 +125,21 @@ func (feed *Feed) Process(pState *state.ProcessState) { // Number entries being processed. errors := 0 processing := 0 + traversed := 0 // Channel for receiving entry results. erChan := make(chan state.EntryResult) - for i, entry := range feed.Entries { + for _, entry := range feed.Entries { e := entry + + // Ignore entry if its title does not matches + // feed's 'title-contains' string. + if len(feed.TitleContains) > 0 && + !e.TitleContains(feed.TitleContains) { + fmt.Printf("[%s][%s]: Skipping '%s'\n", + feed.Id, e.Id, e.Title) + continue + } + // Process entry only if it was not downloaded before. if !pState.DB.Exists(feed.Id, e.Id) { go feed.processEntry(e, erChan) @@ -136,9 +148,10 @@ func (feed *Feed) Process(pState *state.ProcessState) { fmt.Printf("[%s][%s]: Already downloaded '%s' before\n", feed.Id, e.Id, e.Title) } + traversed += 1 // Process only `feed.Last` entries. - if i >= feed.Last-1 { + if traversed >= feed.Last-1 { break } } -- cgit v1.2.3