You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

134 lines
4.3KB

  1. use std::convert::From;
  2. use std::fs::File;
  3. use std::io;
  4. use std::cmp;
  5. use std::io::BufRead;
  6. use std::io::BufReader;
  7. use std::io::Read;
  8. use std::path::PathBuf;
  9. use tantivy;
  10. use tantivy::Index;
  11. use tantivy::IndexWriter;
  12. use tantivy::Document;
  13. use time::PreciseTime;
  14. use clap::ArgMatches;
  15. use chan;
  16. use std::thread;
  17. pub fn run_index_cli(argmatch: &ArgMatches) -> Result<(), String> {
  18. let index_directory = PathBuf::from(argmatch.value_of("index").unwrap());
  19. let document_source = {
  20. match argmatch.value_of("file") {
  21. Some(path) => {
  22. DocumentSource::FromFile(PathBuf::from(path))
  23. }
  24. None => DocumentSource::FromPipe,
  25. }
  26. };
  27. let num_threads = try!(value_t!(argmatch, "num_threads", usize).map_err(|_|format!("Failed to read num_threads argument as an integer.")));
  28. let buffer_size = try!(value_t!(argmatch, "memory_size", usize).map_err(|_|format!("Failed to read the buffer size argument as an integer.")));
  29. let buffer_size_per_thread = buffer_size / num_threads;
  30. run_index(index_directory, document_source, buffer_size_per_thread, num_threads).map_err(|e| format!("Indexing failed : {:?}", e))
  31. }
  32. fn run_index(directory: PathBuf, document_source: DocumentSource, buffer_size_per_thread: usize, num_threads: usize) -> tantivy::Result<()> {
  33. let index = try!(Index::open(&directory));
  34. let schema = index.schema();
  35. let (line_sender, line_receiver) = chan::sync(10_000);
  36. let (doc_sender, doc_receiver) = chan::sync(10_000);
  37. thread::spawn(move || {
  38. let articles = document_source.read().unwrap();
  39. for article_line_res in articles.lines() {
  40. let article_line = article_line_res.unwrap();
  41. line_sender.send(article_line);
  42. }
  43. });
  44. let num_threads_to_parse_json = cmp::max(1, num_threads / 2);
  45. info!("Using {} threads to parse json", num_threads_to_parse_json);
  46. for _ in 0..num_threads_to_parse_json {
  47. let schema_clone = schema.clone();
  48. let doc_sender_clone = doc_sender.clone();
  49. let line_receiver_clone = line_receiver.clone();
  50. thread::spawn(move || {
  51. for article_line in line_receiver_clone {
  52. match schema_clone.parse_document(&article_line) {
  53. Ok(doc) => {
  54. doc_sender_clone.send(doc);
  55. }
  56. Err(err) => {
  57. println!("Failed to add document doc {:?}", err);
  58. }
  59. }
  60. }
  61. });
  62. }
  63. drop(doc_sender);
  64. let mut index_writer = try!(
  65. if num_threads > 0 {
  66. index.writer_with_num_threads(num_threads, buffer_size_per_thread)
  67. }
  68. else {
  69. index.writer(buffer_size_per_thread)
  70. }
  71. );
  72. let index_result = index_documents(&mut index_writer, doc_receiver);
  73. match index_result {
  74. Ok(docstamp) => {
  75. println!("Commit succeed, docstamp at {}", docstamp);
  76. Ok(())
  77. }
  78. Err(e) => {
  79. println!("Error during indexing, rollbacking.");
  80. index_writer.rollback().unwrap();
  81. println!("Rollback succeeded");
  82. Err(e)
  83. }
  84. }
  85. }
  86. fn index_documents(index_writer: &mut IndexWriter, doc_receiver: chan::Receiver<Document>) -> tantivy::Result<u64> {
  87. let group_count = 100_000;
  88. let mut num_docs = 0;
  89. let mut cur = PreciseTime::now();
  90. for doc in doc_receiver {
  91. try!(index_writer.add_document(doc));
  92. if num_docs > 0 && (num_docs % group_count == 0) {
  93. println!("{} Docs", num_docs);
  94. let new = PreciseTime::now();
  95. let elapsed = cur.to(new);
  96. println!("{:?} docs / hour", group_count * 3600 * 1_000_000 as u64 / (elapsed.num_microseconds().unwrap() as u64));
  97. cur = new;
  98. }
  99. num_docs += 1;
  100. }
  101. index_writer.commit()
  102. }
  103. enum DocumentSource {
  104. FromPipe,
  105. FromFile(PathBuf),
  106. }
  107. impl DocumentSource {
  108. fn read(&self,) -> io::Result<BufReader<Box<Read>>> {
  109. Ok(match self {
  110. &DocumentSource::FromPipe => {
  111. BufReader::new(Box::new(io::stdin()))
  112. }
  113. &DocumentSource::FromFile(ref filepath) => {
  114. let read_file = try!(File::open(&filepath));
  115. BufReader::new(Box::new(read_file))
  116. }
  117. })
  118. }
  119. }