note

hzqmwne · Feb 3, 2018 · cc32b44 · cc32b44
1 parent 27a89de
commit cc32b44
Show file tree

Hide file tree

Showing 21 changed files with 1,145 additions and 0 deletions.
diff --git a/README.md b/README.md
@@ -0,0 +1,31 @@
+### SJTU&emsp;SE302&emsp;编译原理与技术(Principle and Technology of Compiler)   
+##### 《Modern Compiler Implementation in C》中的Tiger编译器  
+&emsp;  
+lab1：Straight-line program interpreter（直线式程序解释器）  
+lab2：Lexical Analysis（词法分析）  
+lab3：Syntax Analysis（语法分析）  
+final_tiger：Final Test for lab2 and lab3（对lab2和lab3的深度测试）  
+lab4：Type Checking（类型检查）  
+lab5：Intermediate Code（中间代码生成）  
+lab6：A Workable Tiger Compiler（可用的tiger编译器）  
+&emsp;  
+#### Usage:  
+编译tiger编译器：（需要linux系统并安装有gcc、make、lex、yacc）  
+cd lab6  
+make  
+得到tiger-compiler  
+
+编译tiger程序：  
+./tiger-compiler hello.tig  
+得到hello.tig.s  
+
+链接：（对于64位系统，需要安装32位运行库）  
+gcc -Wl,--wrap,getchar -m32 hello.tig.s runtime.c -o hello.out  
+（如果是32位系统，去掉-m32参数）  
+
+运行：  
+./hello.out  
+&emsp;  
+&emsp;  
+
+更多信息可参考note.txt  
diff --git a/final_tiger/README.md b/final_tiger/README.md
@@ -0,0 +1 @@
+Final Test，仅针对lab2和lab3
diff --git a/lab1/README.md b/lab1/README.md
@@ -0,0 +1,10 @@
+### Lab1 Straight-line program interpreter  
+具体要求参见[lab1.md](./lab1.md)  
+&emsp;  
+&emsp;  
+正式项目开始前的热身。熟悉c语言、递归、树、键值对存储，了解无副作用编程（without side effects，除了初始化以外不使用赋值语句）  
+&emsp;  
+
+需要实现两个函数:  
+int maxargs(A_stm stm)：统计给定语句及其子表达式中参数最多的print语句的参数个数  
+void interp(A_stm stm)：对给定的语句进行解释执行  
diff --git a/lab1/lab1.md b/lab1/lab1.md
@@ -0,0 +1,91 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html>
+<head>
+</head>
+
+<body onload="on_resize()" onclick="on_resize()">
+<div id="header">
+<span class="MainTitle">Compilers</span>
+<span class="SubTitle">2017</span>
+</div>
+
+
+<div id="content">
+
+<h2>Rules</h2>
+	<ol>
+	 <li><font color="red">Prohibit plagiarism!</font></li>
+	 <li>Uploading the lab: you must use <font color="red">`make handin`</font> provided by the Makefile and change the file name with your student ID.</li>
+	 <li>You will get <font color="red">0 point</font> if missing the deadline.</li>
+	</ol>
+
+
+<h2>Lab1 Straight-line program interpreter</h2>
+	<p>
+	<b>Due:</b> Sep 17, 12:00, 2017 <font color="red">At noon!</font></br> 
+	If you have any problem about this lab, please contact TA
+	</p>
+<hr>
+
+
+<h2>Introduction</h2>
+<p>
+In this lab you will implement a simple straight-line program interpreter for the programming language described in <b>chapter 1</b> of your text book. Please read it carefully before you start.
+</p>
+<p>
+This lab serves as an introduction to:
+</p>
+<p>
+	&nbsp&nbsp&nbsp&nbsp<b>environments</b> (symbol tables mapping variable-names to information about the variables)
+</p>
+<p>
+	&nbsp&nbsp&nbsp&nbsp<b>abstract syntax</b> (data structures representing the phrase structure of programs)
+<p>
+<p>
+	&nbsp&nbsp&nbsp&nbsp<b>recursion over tree data structures</b> (which is useful in many parts of a compiler)
+</p>
+
+<p> Please pay attention to the following things: </p>
+	<p>&nbsp&nbsp&nbsp&nbspWrite the interpreter <b>without side effects</b> (Please read your text book carefully)</p>
+
+
+
+<h2>Getting started</h2>
+<p>
+Download the orginal environment <a href="https://ipads.se.sjtu.edu.cn/courses/compilers/handout-2017/lab1.tar.gz">here</a>.
+</p>
+<p>
+Hint: You should first have a quick all the files. You must <font color="red">put all your implementation</font> in the file called `myimpl.c`. Do not modify other files!
+<pre>
+$ tar -xf lab1.tar.gz
+$ cd lab1
+</pre>
+
+</p>
+
+<h2>Testing</h2>
+<p>
+We've provided a script file called <b>gradeMe.sh</b> for testing. After programming, try running:
+<pre>
+$ ./gradeMe.sh
+[^_^]: Pass
+</pre> 
+The smiling face implies that you have passed all the test points. <b>gradeMe.sh</b> compares your output with <b>ref.txt</b>, so if there's something wrong with your answer, you can check the content in <b>ref.txt</b>.
+</p>
+
+<h2>Handin procedure</h2>
+<p>
+Submit your code to TA. You should archive your code as a gzipped tar file with the following command before submission:
+<pre>
+$ cd lab1
+$ make handin 
+$ change the name of tar to: e.g. 5150379000.tar.gz (Do not add other characters!)
+</pre>
+You will receive full credit if your code passes the test in <b>gradeMe.sh</b> and your interpreter is a '<b>without side effect</b>' interpreter. 
+<p></p>
+We will use the timestamp of your <b>last</b> submission for the purpose of calculating late days.
+
+</p>
+</div>
+</body>
+</html>
diff --git a/lab1/note.txt b/lab1/note.txt
@@ -0,0 +1,5 @@
+https://ipads.se.sjtu.edu.cn/courses/compilers/labs/lab1.html
+
+需要实现myimpl.c
+lab1与整个tiger编译器的实现没有关系，但是lab1练习的递归、树、键值对存储将贯穿整个项目。
+关于without side effects，书上有详细的说明
diff --git a/lab2/README.md b/lab2/README.md
@@ -0,0 +1,6 @@
+### Lab2 Lexical Analysis  
+具体要求参见[lab2.md](./lab2.md)   
+&emsp;  
+
+使用lex工具，编写正则表达式对tiger程序进行词法分析  
+注意对字符串和嵌套注释的处理  
diff --git a/lab2/lab2.md b/lab2/lab2.md
@@ -0,0 +1,121 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html>
+<head>
+</head>
+
+<body onload="on_resize()" onclick="on_resize()">
+<div id="header">
+<span class="MainTitle">Compilers</span>
+<span class="SubTitle">2017</span>
+</div>
+
+<div id="content">
+	<h2>Lab2 Lexical Analysis</h2>
+	<p>
+	<b>Due:</b> Oct 7, 12:00 a.m, 2017 <font color="red">At noon!</font></br> 
+	</p>
+<hr>
+
+<h2><font color="red">Update</font></h2>
+<p>
+<a href="https://ipads.se.sjtu.edu.cn/courses/compilers/handout-2017/update.tar.gz">Here</a> are the two new testcases (test50.out and test51.out). Please copy them into directory named "refs".
+</p>
+
+<h2>Introduction</h2>
+<p>
+In this lab you will implement a lexical analyzer for the Tiger language with Lex. Please read <b>chapter 2</b> of your textbook carefully before you start.
+</p>
+
+<h2>Getting started</h2>
+<p>
+Download the orginal environment <a href="https://ipads.se.sjtu.edu.cn/courses/compilers/handout-2017/lab2.tar.gz">here</a>.
+<pre>
+$ tar -xf lab2.tar.gz
+$ cd ~/lab2
+</pre>
+</p>
+
+<h2>Guide</h2>
+<p>
+0. You are only allowed to modify <b>tiger.lex</b> exclude the first several lines.
+</p>
+<p>
+1. This lab is the first step to build our tiger compiler. So you should <b>carefully manage</b> your code including the following labs. For example, using GIT or cloud storage.
+</p>
+
+<p>
+2. We recommend using <b>states</b> to handle comments (please refer to page 32-33 in the text book). Specifically, tiger language allows <b>nested comments</b>.
+This one is legal:
+<pre>
+/* This is comment. /*nested comment*/*/
+</pre>
+While this one is illegal:
+<pre>
+/* This is comment. /*nested comment*/
+</pre>
+</p>
+
+<p>
+3. When handing an "ID", you should use <b>"String"</b> function to allocate the object.
+</p>
+
+<p>
+4. Besides, you should turn in <b>documentation</b> for the following points: 
+</p>
+<p>
+<li> how you handle comments
+<li> how you handle strings
+<li> error handling
+<li> end-of-file handling
+<li> other interesting features of your lexer
+</p>
+<p>
+Send your <b>documentation</b> to <b>TA </b> with the name of `studentID_name_lab2_document.pdf`
+</p>
+
+<h2>Testing</h2>
+<p>
+<b>Warning:</b> Before testing, please make sure that your compiler generates no warnings. Our testing environment might be different from yours, so warnings may be evil sometimes. Watch out! 
+
+<hr>
+We've provided a script file called <b>gradeMe.sh</b> for testing. After programming, try running:
+<pre>
+$ ./gradeMe.sh
+[^_^]: Pass
+</pre> 
+The smiling face implies that you have passed all the test on Tiger code files in <b>/testcases</b>. <b>gradeMe.sh</b> compares your output with the files in <b>/refs</b>, so if there's something wrong with your answer, you can check the content in <b>/refs</b>.
+</p>
+
+<h2>Final Test</h2>
+<p>
+Since our tiger compiler is built step by step, there may exist some hidden bugs in current step.
+Therefore, we provide a final test for uncovering potentail bugs. </br>
+Although this test does not affect your score of this lab, it can help you find bugs as early as possible.
+</p>
+<p>
+In case of some compatible issues, we provide a virutal machine (runs on VMware) for the final test. </br>
+It is <b>not</b> necessary to coding inside the VM (Nevertheless, it has a workable environment.)
+You can download it from <a href="https://jbox.sjtu.edu.cn/l/71KeTu">JBox</a>. VM user: tmac ; password: 123 
+<pre>
+For testing,
+$ copy tiger.lex into VM at first
+$ cp tiger.lex ~/final-tiger
+$ cd final-tiger
+$ ./gradeMe.sh
+</pre>
+</p>
+
+
+<h2>Handin procedure</h2>
+<p>
+Submit your code to TA. You should archive your code as a gzipped tar file with the following command before submission:
+<pre>
+$ cd lab2
+$ make handin
+$ change the name of tar to: e.g. 5150379000.tar.gz (Do not add other characters!)
+</pre>
+You will receive full credit if your code passes the test in <b>gradeMe.sh</b> when running on our machines. We will use the timestamp of your <b>last</b> submission for the purpose of calculating late days.
+</p>
+</div>
+</body>
+</html>
diff --git a/lab2/note.txt b/lab2/note.txt
@@ -0,0 +1,68 @@
+https://ipads.se.sjtu.edu.cn/courses/compilers/labs/lab2.html
+
+这是实现整个编译器的第一步，词法分析。需要完成tiger.lex文件。
+词法分析过程以字节为单位读取源文件，解析出词法单元。
+每次调用词法分析器，返回一个整数作为词法单元的标识。
+
+书上给出了tiger语言的词法单词。大部分都可以直接识别，需要注意的是字符串和嵌套注释
+
+
+嵌套注释：
+lex初始处于INITIAL状态，当遇到"/*"时进入COMMENT状态。在COMMENT状态下，遇到"/*"对计数器加1,遇到"*/"对计数器减1。计数器归零后回到INITIAL状态。
+
+
+字符串：
+tiger语言的字符串支持多种转义字符，包括：（以下摘抄自书）
+\n          换行符
+\t          制表符Tab
+\^c         控制字符c
+\ddd        具有ASCII码ddd（3个十进制数字）的单个字符
+\"          双引号字符（"）
+\\          反斜线字符（\）
+\f___f\     此序列将被忽略，其中f___f代表一个或多个以上的格式化字符
+           （非可打印字符的子集，至少应包含空白符、制表符、换行符、走纸符）
+            组成的序列。（这允许在一行的末尾和下一行的开头各写一个“\”，
+            从而写出一个长度超过一行的长字符串）
+在lab2中，只实现了\n和\t两种转义，其他情况只是简单地忽略“\”字符（这间接实现了\"和\\）。遇到换行也将自动结束字符串。
+具体实现上，当lex遇到双引号（\x22）的时候，调用一个自定义函数逐个字符读取并处理转义，返回识别后的字符串。
+（也可以直接编写正则表达式识别出字符串后再处理转义，但这个正则表达式比较复杂，可读性和调试都比较困难，而且仍需要单独处理转义，故未采用此方法）
+（至于通过正则表达式一次性识别字符串和转义……直接放弃了此方法）
+
+
+====================
+
+
+注意lab2中对长度为0的字符串返回NULL指针而不是空字符串，这只是为了同时通过lab2测试和final test。在lab3及以后，这里返回的是空字符串而非空指针
+测试时发现的另一个问题：C语言printf函数对NULL参数的处理：
+程序一：
+#include <stdio.h>
+int main() {
+    printf("%s\n", NULL);  // ----------> 产生segmentation fault
+    return 0;
+}
+程序二：
+#include <stdio.h>
+int main() {
+    printf("%s", NULL);  // -----> 打印出(NULL)
+    return 0;
+}
+程序三：
+#include <stdio.h>
+int main() {
+    puts(NULL);  // ----------> 产生segmentation fault
+    return 0;
+}
+（以上是在linux平台，gcc 5.4.0 使用 -O0 参数编译运行的结果）
+第一种情况下printf被优化为了puts（尽管用-O0指定了不优化……），但第二中情况没有（可以通过反汇编验证）
+printf对NULL参数会打印出(NULL)（这应该是未定义行为），但puts遇到NULL参数会直接错误
+objdump -d 反汇编的结果对程序一和程序三完全相同
+
+
+======================
+
+
+特别注意，tiger语言允许“\0”作为合法的字符串的一部分，而C语言把“\0”作为字符串的结束符。如果词法分析只是简单的返回指向字符串开头的指针，在后面的编译阶段会丢失“\0”后面的字符串内容（这个bug直到最后生成汇编程序时才发现……）
+对词法分析器的完善参见lab6，添加了对\ddd转义的处理（这样字符串就可以包含任何字符）和对“\0”的特殊处理
+
+lab2提供了final test，用以检验词法分析器能否通过最终的测试。后续lab将要完成的内容在这里以.o文件的形式提供。
+将tiger.lex复制到final_tiger文件夹下（需要先回滚到final_tiger的初始提交状态，因为lab3修改了final_tiger），然后make;./gradeMe.sh
diff --git a/lab3/README.md b/lab3/README.md
@@ -0,0 +1,6 @@
+### Lab3 Syntax Analysis  
+具体要求参见[lab3.md](./lab3.md)   
+&emsp;  
+
+使用yacc，编写文法规则对tiger程序进行语法分析，并生成抽象语法树  
+需要调用lab2写好的词法分析器。将lab2/tiger.lex复制到lab3中