I am trying to write a parser for my language (for learning and fun). The problem is that I don't know how to parse expressions like a--b or a----b. When there are multiple operators made from - character - unary (-x), binary (x-y), pre-decrement (--x, like in C) and post-decrement (x--). Both a--b and a----b should be valid and produce:
a--b -> sub a----b -> sub
+-+-+ +--+--+
a neg decr neg
| | |
b a b
When lexer tokenizes a--b it does not know if it is decrement or minus sign repeated two times, so the parser must find out which one is it.
How could I determine if - is part of decrement operator or just minus sign?
The problem is not really parsing so much as deciding the rules.
Why should a----b be a-- - -b and not a - -(--b)? For that matter, should a---b be a-- - b or a - --b?
And what about a---3 or 3---a? Neither 3-- nor --3 make any sense, so if the criteria were "choose (one of) the sensible interpretations", you'd end up with a-- - 3 and 3 - --a. But even if that were implementable without excess effort, it would place a huge cognitive load on coders and code readers.
Once upon a time, submitting a program for execution was a laborious and sometimes bureaucratic process, and having a run cancelled because the compiler couldn't find the correct interpretation was enormously frustrating. . I still carry the memories of my student days, waiting in a queue to hand my programs to an computer operator and then in another queue to receive the printed results.
So it became momentarily popular to create programming languages which went to heroic lengths to find a valid interpretation of what they were given. But that effort also meant that some bugs passed without error, because the programmer and the programming language having different understandings of what the "natural interpretation" might be.
If you program in C/C++, you may well have at some time written a & 3 == 0 instead of (a & 3) == 0. Fortunately, modern compilers will warn you about this bug, if warnings are enabled. But it's at least reasonable to ask whether the construct should even be permitted. Although it's a little annoying to have to add parentheses and recompile, it's not nearly as frustrating as trying to debug the obscure behaviour which results. Or to have accepted the code in a code review without noticing the subtle error.
These days, the compile / test / edit cycle is much quicker, so there's little excuse fir insisting on clarity. If I were writing a compiler today, I'd probably flag as an error any potentially ambiguous sequence of operator characters. But that might be going too far.
In most languages, a relatively simple rule is used: at each point in the program, the lexical analysis chooses the longest possible token, whether or not it "makes sense". That's what C and C++ do (mostly) and it has the advantage of being easy to implement and also easy to verify. (Even so, in a code review I would insist that a---b be written as a-- -b.)
You could slightly modify this rule so that only the first pair of -- is taken as a token, which would capture some of your desired parses without placing too much load on the code reader.
You could use even more complicated rules, but remember that whatever you implement, you have to document. If it's too hard to document clearly, it's probably unsuitable.
Once you articulate your list of rules, you work on the implementation. In many cases, the simplest is to just try the possibilities in order, with backtracking or in parallel. Or you might be able to just precompute the possible parses.
Alternatively, you could use a GLR or GLL parser generator capable of finding all parses in an ambiguous grammar, and then select the "best" parse based on whatever criteria you prefer.
Related
I am trying to create a simple programming language from scratch (interpreter) but I wonder why I should use a lexer.
For me, it looks like it would be easier to create a parser that directly parses the code. what am I overlooking?
I think you'll agree that most languages (likely including the one you are implementing) have conceptual tokens:
operators, e.g * (usually multiply), '(', ')', ;
keywords, e.g., "IF", "GOTO"
identifiers, e.g. FOO, count, ...
numbers, e.g. 0, -527.23E-41
comments, e.g., /* this text is ignored in your file */
whitespace, e.g., sequences of blanks, tabs and newlines, that are ignored
As a practical matter, it takes a specific chunk of code to scan for/collect the characters that make each individual token. You'll need such a code chunk for each type of token your language has.
If you write a parser without a lexer, at each point where your parser is trying to decide what comes next, you'll have to have ALL the code that recognize the tokens that might occur at that point in the parse. At the next parser point, you'll need all the code to recognize the tokens that are possible there. This gives you an immense amount of code duplication; how many times do you want the code for blanks to occur in your parser?
If you think that's not a good way, the obvious cure to is remove all the duplication: place the code for each token in a subroutine for that token, and at each parser place, call the subroutines for the tokens. At this point, in some sense, you already have a lexer: an isolated collection of code to recognize tokens. You can code perfectly fine recursive descent parsers this way.
The next thing you'll discover is that you call the token subroutines for many of the tokens at each parser point. Even that seems like a lot of work and duplication. So, replace all the calls with a single "GetNextToken" call, that itself invokes the token recognizing code for all tokens, and returns a enum that identifies the specific token encountered. Now your parser starts to look reasonable: at each parser point, it makes one call on GetNextToken, and then branches on enum returned. This is basically the interface that people have standardized on as a "lexer".
One thing you will discover is the token-lexers sometimes have trouble with overlaps; keywords and identifiers usually have this trouble. It is actually easier to merge all the token recognizers into a single finite state machine, which can then distinguish the tokens more easily. This also turns out to be spectacularly fast when processing the programming language source text. Your toy language may never parse more than 100 lines, but real compilers process millions of lines of code a day, and most of that time is spent doing token recognition ("lexing") esp. white space suppression.
You can code this state machine by hand. This isn't hard, but it is rather tedious. Or, you can use a tool like FLEX to do it for you, that's just a matter of convenience. As the number of different kinds of tokens in your language grows, the FLEX solution gets more and more attractive.
TLDR: Your parser is easier to write, and less bulky, if you use a lexer. In addition, if you compile the individual lexemes into a state machine (by hand or using a "lexer generator"), it will run faster and that's important.
Well, for intelligently simplified programing language you can get away without either lexer or parser :-) Not kidding. Look up Forth. You can start with tags here on SO (gforth is GNU's) and then go to the Standard's site which has pointers to a few interpreters, sites and its Glossary.
Then you can check out Win32Forth and that should keep you busy for quite a while :-)
Interpreter also compiles (when you invoke words that switch system to compilation context). All without a distinct parser. Lookahead is actually lookbehind :-) - not kidding. It rarely absorbs one following word (== lookahead is max 1). The "words" (aka tokens) are at the same time keywords and variable names and they all live in a Dictionary. There's a whole online book at that site (plus pdf).
Control structures are also just words (they compile a few addresses and jumps on the fly).
You can find old Journals there as well, covering a wide spectrum from machine code generation to object oriented extensions. Yes still without parser - believe it or not.
There used to be more sophisticated (commercial) Forth systems which were reducing words to machine call instructions with immediate addressing (makes the engine run 2-4 times faster) but even plain interpreters were always considered to be fast. One is apparently still active - SwiftForth, but don't expect any freebies there.
There's one Forth on GitHub CiForth which is quite spartanic but has builds and releases for Win, Linux and Mac, 32 and 64 so you can just download and run. Claims to have a 16-bit build as well :-) For embedded systems I suppose.
I have a project that doesn't have -spec or -type in code, currently dialyzer can find some warnings, most of them are in machine generated codes.
Will adding type specs to code make dialyzer find more errors?
Off the topic, is there any tool to check whether the specs been violated or not?
Adding typespecs will dramatically improve the accuracy of Dialyzer.
Because Erlang is a dynamic language Dialyzer must default to a rather broad interpretation of types unless you give it hints to narrow the "success" typing that it will go by. Think of it like giving Dialyzer a filter by which it can transform a set of possible successes to a subset of explicit types that should ever work.
This is not the same as Haskell, where the default assumption is failure and all code must be written with successful typing to be compiled at all -- Dialyzer must default to assume success unless it knows for sure that a type is going to fail.
Typespecs are the main part of this, but Dialyzer also checks guards, so a function like
increment(A) -> A + 1.
Is not the same as
increment(A) when A > 100 -> A + 1.
Though both may be typed as
-spec increment(integer()) -> integer().
Most of the time you only care about integer values being integer(), pos_integer(), neg_integer(), or non_neg_integer(), but occasionally you need an arbitrary range bounded only on one side -- and the type language has no way to represent this currently (though personally I would like to see a declaration of 100..infinity work as expected).
The unbounded-range of when A > 100 requires a guard, but a bounded range like when A > 100 and A < 201 could be represented in the typespec alone:
-spec increment(101..200) -> pos_integer().
increment(A) -> %stuff.
Guards are fast with the exception of calling length/1 (which you should probably never actually need in a guard), so don't worry with the performance overhead until you actually know and can demonstrate that you have a performance problem that comes from guards. Using guards and typespecs to constrain Dialyzer is extremely useful. It is also very useful as documentation for yourself and especially if you use edoc, as the typespec will be shown there, making APIs less mysterious and easy to toy with at a glance.
There is some interesting literature on the use of Dialyzer in existing codebases. A well-documented experience is here: Gradual Typing of Erlang Programs: A Wrangler Experience. (Unfortunately some of the other links I learned a lot from previously have disappeared or moved. (!.!) A careful read of the Wrangler paper, skimming over the User's Guide and man page, playing with Dialyzer, and some prior experience in a type system like Haskell's will more than prepare you for getting a lot of mileage out of Dialyzer, though.)
[On a side note, I've spoken with a few folks before about specifying "pure" functions that could be guaranteed as strongly typed either with a notation or by using a different definition syntax (maybe Prolog's :- instead of Erlang's ->... or something), but though that would be cool, and it is very possible even now to concentrate side-effects in a tiny part of the program and pass all results back in a tuple of {Results, SideEffectsTODO}, this is simply not a pressing need and Erlang works pretty darn well as-is. But Dialyzer is indeed very helpful for showing you where you've lost track of yourself!]
I'm writing a program that takes in input a straight play in a custom format and then performs some analysis on it (like number of lines and words for each character). It's just for fun, and a pretext for learning cool stuff.
The first step in that process is writing a parser for that format. It goes :
####Play
###Act I
##Scene 1
CHARACTER 1. Line 1, he's saying some stuff.
#Comment, stage direction
CHARACTER 2, doing some stuff. Line 2, she's saying some stuff too.
It's quite a simple format. I read extensively about basic parser stuff like CFG, so I am now ready to get some work done.
I have written my grammar in EBNF and started playing with flex/bison but it raises some questions :
Is flex/bison too much for such a simple parser ? Should I just write it myself as described here : Is there an alternative for flex/bison that is usable on 8-bit embedded systems? ?
What is good practice regarding the respective tasks of the tokenizer and the parser itself ? There is never a single solution, and for such a simple language they often overlap. This is especially true for flex/bison, where flex can perform some intense stuff with regex matching. For example, should "#" be a token ? Should "####" be a token too ? Should I create types that carry semantic information so I can directly identify for example a character ? Or should I just process it with flex the simplest way then let the grammar defined in bison decide what is what ?
With flex/bison, does it makes sense to perform the analysis while parsing or is it more elegant to parse first, then operate on the file again with some other tool ?
This got me really confused. I am looking for an elegant, perhaps simple solution. Any guideline ?
By the way, about the programing language, I don't care much. For now I am using C because of flex/bison but feel free to advise me on anything more practical as long as it is a widely used language.
It's very difficult to answer those questions without knowing what your parsing expectations are. That is, an example of a few lines of text does not provide a clear understanding of what the intended parse is; what the lexical and syntactic units are; what relationships you would like to extract; and so on.
However, a rough guess might be that you intend to produce a nested parse, where ##{i} indicates the nesting level (inversely), with i≥1, since a single # is not structural. That violates one principle of language design ("don't make the user count things which the computer could count more accurately"), which might suggest a structure more like:
#play {
#act {
#scene {
#location: Elsinore. A platform before the castle.
#direction: FRANCISCO at his post. Enter to him BERNARDO
BERNARDO: Who's there?
FRANCISCO: Nay, answer me: stand, and unfold yourself.
BERNARDO: Long live the king!
FRANCISCO: Bernardo?
or even something XML-like. But that would be a different language :)
The problem with parsing either of these with a classic scanner/parser combination is that the lexical structure is inconsistent; the first token on a line is special, but most of the file consists of unparsed text. That will almost inevitably lead to spreading syntactic information between the scanner and the parser, because the scanner needs to know the syntactic context in order to decide whether or not it is scanning raw text.
You might be able to avoid that issue. For example, you might require that a continuation line start with whitespace, so that every line not otherwise marked with #'s starts with the name of a character. That would be more reliable than recognizing a dialogue line just because it starts with the name of a character and a period, since it is quite possible for a character's name to be used in dialogue, even at the end of a sentence (which consequently might be the first word in a continuation line.)
If you do intend for dialogue lines to be distinguished by the fact that they start with a character name and some punctuation then you will definitely have to give the scanner access to the character list (as a sort of symbol table), which is a well-known but not particularly respected hack.
Consider the above a reflection about your second question ("What are the roles of the scanner and the parser?"), which does not qualify as an answer but hopefully is at least food for thought. As to your other questions, and recognizing that all of this is opinionated:
Is flex/bison too much for such a simple parser ? Should I just write it myself...
The fact that flex and bison are (potentially) more powerful than necessary to parse a particular language is a red herring. C is more powerful than necessary to write a factorial function -- you could easily do it in assembler -- but writing a factorial function is a good exercise in learning C. Similarly, if you want to learn how to write parsers, it's a good idea to start with a simple language; obviously, that's not going to exercise every option in the parser/scanner generators, but it will get you started. The question really is whether the language you're designing is appropriate for this style of parsing, not whether it is too simple.
With flex/bison, does it makes sense to perform the analysis while parsing or is it more elegant to parse first, then operate on the file again with some other tool?
Either can be elegant, or disastrous; elegance has more to do with how you structure your thinking about the problem at hand. Having said that, it is often better to build a semantic structure (commonly referred to as an AST -- abstract syntax tree) during the parse phase and then analyse that structure using other functions.
Rescanning the input file is very unlikely to be either elegant or effective.
I am trying to create a simple script for a resource API. I have a resource API mainly creates game resources in a structured manner. What I want is dealing with this API without creating c++ programs each time I want a resource. So we (me and my instructor from uni) decided to create a simple script to create/edit resource files without compiling every time. There are also some other irrelevant factors that I need a command line interface rather than a GUI program.
Anyway, here is script sample:
<path>.<command> -<options>
/Graphics[3].add "blabla.png"
I didn't design this script language, the owner of API did. The part before '.' as you can guess is the path and part after '.' is actual command and some options, flags etc. As a first step, I tried to create grammar of left part because I thought I could use it while searching info about lexical analyzers and parser. The problem is I am inexperienced when it comes to parsing and programming languages and I am not sure if it's correct or not. Here is some more examples and grammar of left side.
dir -> '/' | '/' path
path -> object '/' path | object
object -> number | string '[' number ']'
Notation if this grammar can be a mess, I don't know. There is 5 different possibilities, they are:
String
"String"
Number
String[Number]
"String"[Number]
It has to start with '/' symbol and if it's the only symbol, I will accept it as Root.
Now my problem is how can I lexically analyze this script? Is there a special method? What should my lexical analyzer do and do not(I read some lexical analysers also do syntactic analysis up to a point). Do you think grammar, etc. is technically appropriate? What kind of parsing method I should use(Recursive Descent, LL etc.)? I am trying to make it technically appropriate piece of work. It's not commercial so I have time thus I can learn lexical analysis and parsing better. I don't want to use a parser library.
What should my lexical analyzer do and not do?
It should:
recognize tokens
ignore ignorable whitespace and comments (if there are such things)
optionally, keep track of source location in order to produce meaningful error messages.
It should not attempt to parse the input, although that will be very tempting with such a simple language.
From what I can see, you have the following tokens:
punctuation: /, ., linear-white-space, new-line
numbers
unquoted strings (often called "atoms" or "ids")
quoted strings (possibly the same token type as unquoted strings)
I'm not sure what the syntax for -options is, but that might include more possibilities.
Choosing to return linear-white-space (that is, a sequence consisting only of tabs and spaces) as a token is somewhat questionable; it complicates the grammar considerably, particularly since there are probably places where white-space is ignorable, such as the beginning and end of a line. But I have the intuition that you do not want to allow whitespace inside of a path and that you plan to require it between the command name and its arguments. That is, you want to prohibit:
/left /right[3] .whimper "hello, world"
/left/right[3].whimper"hello, world"
But maybe I'm wrong. Maybe you're happy to accept both. That would be simpler, because if you accept both, then you can just ignore linear-whitespace altogether.
By the way, experience has shown that using new-line to separate commands can be awkward; sooner or later you will need to break a command into two lines in order to avoid having to buy an extra monitor to see the entire line. The convention (used by bash and the C preprocessor, amongst others) of putting a \ as the last character on a line to be continued is possible, but can lead to annoying bugs (like having an invisible space following the \ and thus preventing it from really continuing the line).
From here down is 100% personal opinion, offered for free. So take it for what its worth.
I am trying to make it technically appropriate piece of work. It's not commercial so I have time thus I can learn lexical analysis and parsing better. I don't want to use a parser library.
There is a contradiction here, in my opinion. Or perhaps two contradictions.
A technically appropriate piece of work would use standard tools; at least a lexical generator and probably a parser generator. It would do that because, properly used, the lexical and grammatical descriptions provided to the tools document precisely the actual language, and the tools guarantee that the desired language is what is actually recognized. Writing ad hoc code, even simple lexical recognizers and recursive descent parsers, for all that it can be elegant, is less self-documenting, less maintainable, and provides fewer guarantees of correctness. Consequently, best practice is "use standard tools".
Secondly, I disagree with your instructor (if I understand their proposal correctly, based on your comments) that writing ad hoc lexers and parsers aids in understanding lexical and parsing theory. In fact, it may be counterproductive. Bottom-up parsing, which is incredibly elegant both theoretically and practically, is almost impossible to write by hand and totally impossible to read. Consequently, many programmers prefer to use recursive-descent or Pratt parsers, because they understand the code. However, such parsers are not as powerful as a bottom-up parser (particularly GLR or Earley parsers, which are fully general), and their use leads to unnecessary grammatical compromises.
You don't need to write a regular expression library to understand regular expressions. The libraries abstract away the awkward implementation details (and there are lots of them, and they really are awkward) and let you concentrate on the essence of creating and using regular expressions.
In the same way, you do not need to write a compiler in order to understand how to program in C. After you have a good basis in C, you can improve your understanding (maybe) by understanding how it translates into machine code, but unless you plan a career in compiler writing, knowing the details of obscure optimization algorithms are not going to make you a better programmer. Or, at least, they're not first on your agenda.
Similarly, once you really understand regular expressions, you might find writing a library interesting. Or not -- you might find it incredibly frustrating and give up after a couple of months of hard work. Either way, you will appreciate existing libraries more. But learn to use the existing libraries first.
And the same with parser generators. If you want to learn how to translate an idea for a programming language into something precise and implementable, learn how to use a parser generator. Only after you have mastered the theory of parsing should you even think of focusing on low-level implementations.
So I'm doing a Parser, where I favor flexibility over speed, and I want it to be easy to write grammars for, e.g. no tricky workaround rules (fake rules to solve conflicts etc, like you have to do in yacc/bison etc.)
There's a hand-coded Lexer with a fixed set of tokens (e.g. PLUS, DECIMAL, STRING_LIT, NAME, and so on) right now there are three types of rules:
TokenRule: matches a particular token
SequenceRule: matches an ordered list of rules
GroupRule: matches any rule from a list
For example, let's say we have the TokenRule 'varAccess', which matches token NAME (roughly /[A-Za-z][A-Za-z0-9_]*/), and the SequenceRule 'assignment', which matches [expression, TokenRule(PLUS), expression].
Expression is a GroupRule matching either 'assignment' or 'varAccess' (the actual ruleset I'm testing with is a bit more complete, but that'll do for the example)
But now let's say I want to parse
var1 = var2
And let's say the Parser begins with rule Expression (the order in which they are defined shouldn't matter - priorities will be solved later). And let's say the GroupRule expression will first try 'assignment'. Then since 'expression' is the first rule to be matched in 'assignment', it will try to parse an expression again, and so on until the stack is filled up and the computer - as expected - simply gives up in a sparkly segfault.
So what I did is - SequenceRules add themselves as 'leafs' to their first rule, and become non-roôt rules. Root rules are rules that the parser will first try. When one of those is applied and matches, it tries to subapply each of its leafs, one by one, until one matches. Then it tries the leafs of the matching leaf, and so on, until nothing matches anymore.
So that it can parse expressions like
var1 = var2 = var3 = var4
Just right =) Now the interesting stuff. This code:
var1 = (var2 + var3)
Won't parse. What happens is, var1 get parsed (varAccess), assign is sub-applied, it looks for an expression, tries 'parenthesis', begins, looks for an expression after the '(', finds var2, and then chokes on the '+' because it was expecting a ')'.
Why doesn't it match the 'var2 + var3' ? (and yes, there's an 'add' SequenceRule, before you ask). Because 'add' isn't a root rule (to avoid infinite recursion with the parse-expresssion-beginning-with-expression-etc.) and that leafs aren't tested in SequenceRules otherwise it would parse things like
reader readLine() println()
as
reader (readLine() println())
(e.g. '1 = 3' is the expression expected by add, the leaf of varAccess a)
whereas we'd like it to be left-associative, e.g. parsing as
(reader readLine()) println()
So anyway, now we've got this problem that we should be able to parse expression such as '1 + 2' within SequenceRules. What to do? Add a special case that when SequenceRules begin with a TokenRule, then the GroupRules it contains are tested for leafs? Would that even make sense outside that particular example? Or should one be able to specify in each element of a SequenceRule if it should be tested for leafs or not? Tell me what you think (other than throw away the whole system - that'll probably happen in a few months anyway)
P.S: Please, pretty please, don't answer something like "go read this 400pages book or you don't even deserve our time" If you feel the need to - just refrain yourself and go bash on reddit. Okay? Thanks in advance.
LL(k) parsers (top down recursive, whether automated or written by hand) require refactoring of your grammar to avoid left recursion, and often require special specifications of lookahead (e.g. ANTLR) to be able to handle k-token lookahead. Since grammars are complex, you get to discover k by experimenting, which is exactly the thing you wish to avoid.
YACC/LALR(1) grammars aviod the problem of left recursion, which is a big step forward. The bad news is that there are no real programming langauges (other than Wirth's original PASCAL) that are LALR(1). Therefore you get to hack your grammar to change it from LR(k) to LALR(1), again forcing you to suffer the experiments that expose the strange cases, and hacking the grammar reduction logic to try to handle K-lookaheads when the parser generators (YACC, BISON, ... you name it) produce 1-lookahead parsers.
GLR parsers (http://en.wikipedia.org/wiki/GLR_parser) allow you to avoid almost all of this nonsense. If you can write a context free parser, under most practical circumstances, a GLR parser will parse it without further effort. That's an enormous relief when you try to write arbitrary grammars. And a really good GLR parser will directly produce a tree.
BISON has been enhanced to do GLR parsing, sort of. You still have to write complicated logic to produce your desired AST, and you have to worry about how to handle failed parsers and cleaning up/deleting their corresponding (failed) trees. The DMS Software Reengineering Tookit provides standard GLR parsers for any context free grammar, and automatically builds ASTs without any additional effort on your part; ambiguous trees are automatically constructed and can be cleaned up by post-parsing semantic analyis. We've used this to do define 30+ language grammars including C, including C++ (which is widely thought to be hard to parse [and it is almost impossible to parse with YACC] but is straightforward with real GLR); see C+++ front end parser and AST builder based on DMS.
Bottom line: if you want to write grammar rules in a straightforward way, and get a parser to process them, use GLR parsing technology. Bison almost works. DMs really works.
My favourite parsing technique is to create recursive-descent (RD) parser from a PEG grammar specification. They are usually very fast, simple, and flexible. One nice advantage is you don't have to worry about separate tokenization passes, and worrying about squeezing the grammar into some LALR form is non-existent. Some PEG libraries are listed [here][1].
Sorry, I know this falls into throw away the system, but you are barely out of the gate with your problem and switching to a PEG RD parser, would just eliminate your headaches now.