Finch/src/linker.fch

// From LLVM IR to a program: compile the IR, add the runtime (compiled once and kept), link the C libraries
// and C files the program asks for, and turn the linker's complaints into something a person can act on.
import platform
import cimport
import host

// 32-bit FNV-1a, as hex: names the cached runtime object after what it was built from
fn fingerprint(str text) -> str {
    u32 h = 2166136261
    for c in text {
        h = (h ^ u32(c)) * 16777619
    }
    digits := "0123456789abcdef"
    []str parts
    for k in 0..8 {
        parts.push(str(digits[int((h >> u32(28 - 4 * k)) & 15)]))
    }
    return parts.join("")
}

fn quote(str a) -> str {
    return host.quote(a)
}

// "-lname", or whatever pkg-config says the library needs
fn libFlags(str name, platform.Target tg) -> str {
    if name.ends_with(".framework") {
        return "-framework " + quote(name.sub(0, name.len - 10))         // macOS: link "Cocoa.framework"
    }
    if !tg.cross {
        tmp := host.tempDir() + "/finch_pkg_flags_" + str(host.pid()) + ".txt"
        code := host.run("pkg-config --libs " + quote(name) + " > " + quote(tmp) + " 2> " + host.nullDevice())
        if code == 0 && file_exists(tmp) {
            flags := (read_file(tmp) or "").trim()
            host.remove(tmp)
            if flags.len > 0 {
                return flags
            }
        }
    }
    return quote("-l" + name)
}

// "GLFW/glfw3.h" -> "glfw": a decent first guess for the library's name
fn guessLib(str header) -> str {
    name := header
    slash := name.find("/")
    while slash >= 0 {
        name = name.sub(slash + 1, name.len)
        slash = name.find("/")
    }
    dot := name.len
    for i in 0..name.len {
        if name[i] == '.' {
            dot = i
        }
    }
    name = name.sub(0, dot)
    while name.len > 0 && name[name.len - 1] >= '0' && name[name.len - 1] <= '9' {
        name = name.sub(0, name.len - 1)
    }
    return name.lower()
}

struct Names {
    []str items
}

fn Names.add(str x) {
    if x.len > 0 && !self.items.contains(x) {
        self.items.push(x)
    }
}

// the text after `marker` up to the next one of `stops`
fn between(str line, str marker, str stops) -> str {
    a := line.find(marker)
    if a < 0 {
        return ""
    }
    rest := line.sub(a + marker.len, line.len)
    n := 0
    while n < rest.len && !stops.contains(str(rest[n])) {
        n += 1
    }
    return rest.sub(0, n)
}

// The linker speaks in riddles; say what is missing and what to write.
fn linkFailed(str output, cimport.CImports ci, platform.Target tg) -> int {
    missing := Names()
    libs := Names()
    for line in output.split("\n") {
        l := line.trim()
        if l.contains("undefined reference to `") {
            missing.add(between(l, "undefined reference to `", "'"))
        } else if l.contains("undefined symbol: ") {
            missing.add(between(l, "undefined symbol: ", " \n"))
        } else if l.contains("unresolved external symbol ") {
            missing.add(between(l, "unresolved external symbol ", " \n\r"))
        } else if l.contains("cannot find -l") {
            libs.add(between(l, "cannot find -l", ": \n\r"))
        } else if l.contains("library not found for -l") {
            libs.add(between(l, "library not found for -l", ": \n\r"))
        } else if l.contains("ld: library '") {
            libs.add(between(l, "ld: library '", "'"))
        } else if l.contains("framework '") && l.contains("not found") {
            libs.add(between(l, "framework '", "'") + ".framework")
        } else if l.starts_with("\"") && l.contains("\", referenced from:") {
            // Apple ld: "_name", referenced from:
            name := l.sub(1, l.find("\", referenced from:"))
            if tg.darwin && name.starts_with("_") {
                name = name.sub(1, name.len)
            }
            missing.add(name)
        }
    }
    if missing.items.len == 0 && libs.items.len == 0 {
        print(output + "error: linking failed")
        return 1
    }
    for lib in libs.items {
        print("error: the library '" + lib + "' wasn't found. Is it installed? (on Nix: add it to shell.nix)")
    }
    // the missing functions, grouped by the header they came from
    []str headers
    []Names byHeader
    for name in missing.items {
        header := ""
        if ci.fns.has(name) {
            header = ci.fns[name].header
        }
        k := headers.find(header)
        if k < 0 {
            headers.push(header)
            byHeader.push(Names())
            k = headers.len - 1
        }
        byHeader[k].add(name)
    }
    for k in 0..headers.len {
        names := byHeader[k].items
        []str shown
        for i in 0..names.len {
            if i < 3 {
                shown.push(names[i])
            }
        }
        list := shown.join(", ")
        if names.len > 3 {
            list = list + " and " + str(names.len - 3) + " more"
        }
        if headers[k].len == 0 {
            print("error: no code found for " + list)
            continue
        }
        verb := "come"
        if names.len == 1 {
            verb = "comes"
        }
        print("error: " + list + " " + verb + " from \"" + headers[k] + "\", but its library isn't linked")
        print("  the header only says the functions exist; their code lives in a library.")
        print("  add this at the top of your file (with the library's real name):")
        print("")
        print("      link \"" + guessLib(headers[k]) + "\"")
        print("")
    }
    return 1
}

// The runtime, compiled once per system and compiler and kept in the cache folder.
fn runtimeObject(str rt, platform.Target tg, str cc) -> str! {
    source := try read_file(rt)
    dir := host.cacheDir()
    obj := dir + "/rt-" + fingerprint(source + cc + tg.triple + "2.5") + ".o"
    if file_exists(obj) {
        return obj
    }
    host.makeDirs(dir)
    part := obj + ".part" + str(host.pid())
    pic := " -fPIC"
    if tg.windows {
        pic = ""
    }
    log := part + ".log"
    code := host.run(cc + " -O2" + pic + " -c " + quote(rt) + " -o " + quote(part) + " > " + quote(log) + " 2>&1")
    if code != 0 {
        text := read_file(log) or ""
        host.remove(part)
        host.remove(log)
        return error(text + "error: couldn't build the Finch runtime with '" + cc + "' (is a C compiler installed?)")
    }
    host.remove(log)
    host.move(part, obj)
    return obj
}

struct Build {
    str out             // the program
    str ir              // its LLVM IR, as text
    str rt              // the runtime's source file
    str opt             // -O2 or -O0
    str extra           // -g, -fPIC ...
    bool shared
    []str links         // what the program asked to link: a library name, a .c file (with its folder), a .o or .a
}

// IR -> program. 0 if it worked.
fn build(Build b, platform.Target tg, cimport.CImports ci) -> int {
    clang := getenv("FINCH_CLANG")
    if clang.len == 0 {
        clang = "clang"
    }
    // the C compiler of the runtime and the linker: the target's own when building for another system
    cc := clang
    if tg.cross {
        cc = tg.cc
    } else if getenv("FINCH_CC").len > 0 {
        cc = getenv("FINCH_CC")
    }
    rtobj := runtimeObject(b.rt, tg, cc) or {
        print(err)
        return 1
    }
    ll := b.out + ".ll"
    obj := b.out + ".o"
    write_file(ll, b.ir)
    target := ""
    if tg.cross || tg.windows {
        target = " --target=" + tg.triple
    }
    log := b.out + ".log"
    code := host.run(clang + target + " " + b.opt + b.extra + " -w -c " + quote(ll) + " -o " + quote(obj) + " > " + quote(log) + " 2>&1")
    host.remove(ll)
    if code != 0 {
        print(read_file(log) or "")
        print("error: clang couldn't compile the program's IR")
        host.remove(obj)
        host.remove(log)
        return 1
    }
    cmd := cc + " " + quote(obj) + " " + quote(rtobj) + " -o " + quote(b.out)
    if !tg.msvc {
        cmd = cmd + " -lm"
    }
    if getenv("FINCH_LDFLAGS").len > 0 {
        cmd = cmd + " " + getenv("FINCH_LDFLAGS")          // more flags for the linker, like -L/where/libclang/is
    }
    if b.shared {
        if tg.darwin {
            cmd = cmd + " -dynamiclib"
        } else {
            cmd = cmd + " -shared"
        }
    }
    []str temps
    for l in b.links {
        if l.ends_with(".c") {
            // the program's own C file, built along with it
            o := b.out + "." + str(temps.len) + ".o"
            pic := " -fPIC"
            if tg.windows {
                pic = ""
            }
            c := host.run(cc + " -O2" + pic + " -c " + quote(l) + " -o " + quote(o) + " > " + quote(log) + " 2>&1")
            if c != 0 {
                print((read_file(log) or "") + "error: couldn't compile " + l)
                host.remove(obj)
                host.remove(log)
                host.remove(o)
                return 1
            }
            temps.push(o)
            cmd = cmd + " " + quote(o)
        } else if l.ends_with(".o") || l.ends_with(".a") {
            cmd = cmd + " " + quote(l)
        } else {
            cmd = cmd + " " + libFlags(l, tg)
        }
    }
    status := host.run(cmd + " > " + quote(log) + " 2>&1")
    output := read_file(log) or ""
    host.remove(obj)
    host.remove(log)
    for t in temps {
        host.remove(t)
    }
    if status == 127 {
        print("error: no C compiler found to link with ('" + cc + "'; install one, or set FINCH_CC)")
        return 1
    }
    if status != 0 {
        return linkFailed(output, ci, tg)
    }
    if output.len > 0 {
        print(output)           // warnings, if any
    }
    return 0
}