first

src/main.rs

@@ -1,3 +1,188 @@
-fn main() {
-    println!("Hello, world!");
+extern crate encoding;
+extern crate json;
+extern crate textstream;
+extern crate zip;
+
+use encoding::all::KOI8_R;
+use encoding::DecoderTrap;
+use std::fs;
+use std::io::BufReader;
+use std::io::Read;
+use std::path::PathBuf;
+use textstream::TextReader;
+
+#[derive(Debug, Clone, Copy)]
+enum KeywordType {
+	Global,
+	QuestionPre,
+	QuestionStart,
+	QuestionContent,
+}
+
+fn keyword_type(pattern: &str) -> KeywordType {
+	match pattern {
+		"Чемпионат:" | "URL:" | "Дата:" | "Редактор:" | "Вид:" => {
+			KeywordType::Global
+		}
+		"Тур:" => KeywordType::QuestionPre,
+		"Вопрос " => KeywordType::QuestionStart,
+		_ => KeywordType::QuestionContent,
+	}
+}
+
+struct Context {
+	// global output value
+	data: json::JsonValue,
+	// temp questions array
+	questions: json::JsonValue,
+	cur_keyword_type: Option<KeywordType>,
+	// temp question value
+	cur_question: json::JsonValue,
+	// temp value for pre'question fields
+	cur_question_pre: json::JsonValue,
+	// curent json key
+	cur_tag: String,
+	// current json value
+	cur_content: Vec<String>,
+	// need to push temp question value if true
+	have_new_question: bool,
+	// prev. keyword type
+	last_keyword_type: Option<KeywordType>,
+	// prev. json key (used for store acummulated content when new keyword readed)
+	last_tag: String,
+}
+
+fn parse_file<R: Read>(file: R) -> Result<json::JsonValue, Box<std::error::Error>> {
+	let buf = BufReader::new(file);
+	let reader = TextReader::new(buf, KOI8_R, DecoderTrap::Ignore);
+
+	let patterns = vec![
+		"Чемпионат:",
+		"URL:",
+		"Дата:",
+		"Редактор:",
+		"Вид:",
+		"Тур:",
+		"Вопрос ",
+		"Ответ:",
+		"Зачет:",
+		"Источник:",
+		"Автор:",
+		"Комментарий:",
+	];
+	// init context
+	let mut context = Context {
+		data: json::JsonValue::new_object(),
+		questions: json::JsonValue::new_array(),
+		cur_keyword_type: None,
+		cur_question: json::JsonValue::new_object(),
+		cur_question_pre: json::JsonValue::new_object(),
+		cur_tag: String::new(),
+		cur_content: Vec::<String>::new(),
+		have_new_question: false,
+		last_keyword_type: None,
+		last_tag: String::new(),
+	};
+	let mut ctx = &mut context;
+
+	for line in reader.lines() {
+		// ignore empty lines
+		let line_str = String::from(line.unwrap().trim());
+		let line_s = &line_str;
+		if 0 == line_s.len() {
+			continue;
+		}
+		// find keywords
+		match patterns
+			.iter()
+			.find(|&&pattern| line_s.starts_with(pattern) && line_s.ends_with(":"))
+		{
+			Some(pattern) => {
+				ctx.last_keyword_type = ctx.cur_keyword_type;
+				ctx.last_tag = ctx.cur_tag.clone();
+				ctx.cur_keyword_type = Some(keyword_type(&pattern));
+				ctx.cur_tag = pattern.replace(" ", "").replace(":", "");
+
+				// remember question id
+				match ctx.cur_keyword_type {
+					Some(KeywordType::QuestionStart) => {
+						ctx.cur_question_pre["id"] = line_s.replace(":", "").as_str().into()
+					}
+					_ => (),
+				};
+
+				// apply accumulated content when new keyword found
+				match ctx.last_keyword_type {
+					Some(KeywordType::Global) => {
+						ctx.data[&ctx.last_tag] = ctx.cur_content.join("\n").into()
+					}
+					Some(KeywordType::QuestionPre) => {
+						ctx.cur_question_pre[&ctx.last_tag] = ctx.cur_content.join("\n").into();
+					}
+					Some(KeywordType::QuestionContent) => {
+						ctx.cur_question[&ctx.last_tag] = ctx.cur_content.join("\n").into();
+					}
+					Some(KeywordType::QuestionStart) => {
+						// store prev question before reading new
+						if ctx.have_new_question {
+							ctx.questions.push(ctx.cur_question.clone()).unwrap();
+						}
+						// prepare for read new question data with cur_question_pre values
+						ctx.cur_question = ctx.cur_question_pre.clone();
+						ctx.cur_question_pre = json::JsonValue::new_object();
+						ctx.cur_question[&ctx.last_tag] = ctx.cur_content.join("\n").into();
+						ctx.have_new_question = true;
+					}
+					None => (),
+				};
+				// clear content
+				ctx.cur_content.clear();
+			}
+			None => {
+				// accumulate content if line is not a keyword
+				ctx.cur_content.push(String::from(line_s));
+			}
+		}
+	}
+
+	// finish reading last question
+	if ctx.have_new_question && !ctx.cur_content.is_empty() {
+		ctx.cur_question[&ctx.cur_tag] = ctx.cur_content.join("\n").as_str().into();
+		ctx.questions.push(ctx.cur_question.clone()).unwrap();
+		ctx.have_new_question = false;
+	}
+
+	ctx.data["Вопросы"] = ctx.questions.clone();
+	Ok(ctx.data.clone())
+}
+
+fn main() -> Result<(), Box<std::error::Error>> {
+	let fname = std::path::Path::new("./baza.zip");
+	let zip_file = fs::File::open(&fname)?;
+	let zip_reader = BufReader::new(zip_file);
+
+	let mut archive = zip::ZipArchive::new(zip_reader)?;
+
+	for i in 0..archive.len() {
+		let file = archive.by_index(i)?;
+		// FIXME
+		//if ! file.is_file() {
+		//    continue;
+		//}
+
+		let name = file.sanitized_name();
+		println!("{}", name.as_path().display());
+		let data: json::JsonValue = parse_file(file)?;
+		let mut outfilename = PathBuf::from("./json");
+		outfilename.push(name);
+		outfilename.set_extension("json");
+
+		let mut outfile = fs::File::create(outfilename)?;
+
+		data.write_pretty(&mut outfile, 1)?;
+		//data.write(&mut outfile)?;
+		//debug
+		//break;
+	}
+	Ok(())
 }