Slide 1
Slide 1 text
໌͔Β graphlibɺΈΜͳͰ͓͏
Hayao Suzuki
PyCon JP 2025 at International Conference Center Hiroshima
September 26, 2025
Slide 2
Slide 2 text
Share it
GitHub
› https://github.com/HayaoSuzuki/pyconjp2025
Hashtag
› #pyconjp #PyConJP2025 ɹ#pyconjp_3
2 / 34
Slide 3
Slide 3 text
Who am I ?
͓લ୭Α
Name Hayao Suzukiʢླɹॣʣ
/////////
Twitter X @CardinalXaro
Work ιϑτΣΞΤϯδχΞ at ౦ژΨεגࣜձࣾ
౦ژΨεגࣜձࣾʹ͍ͭͯ
› ҰݝʹࢢΨεɾిؾͳͲͷΤωϧΪʔΛڙڅ͢Δձࣾ
› ౦ژΨε PyCon JP 2025 ͷ Gold εϙϯαʔͰ͢
› ιϑτΣΞΤϯδχΞΛઈࢍืूத
https://www.tokyo-gas-recruit.com/career/
3 / 34
Slide 4
Slide 4 text
Who am I ?
༁
› Effective Python ୈ 3 ൛ (O’Reilly Japan) New!
› ϋΠύʔϞμϯ Python(O’Reilly Japan)
› Python Distilled(O’Reilly Japan)
༁ɾम
› ϩόετ Python(O’Reilly Japan)
› ೖ Python 3 ୈ 2 ൛ (O’Reilly Japan)
› Python ΫΠοΫϦϑΝϨϯε ୈ 4 ൛ (O’Reilly Japan)
4 / 34
Slide 5
Slide 5 text
Who am I ?
աڈͷൃදʢൈਮʣ
› Let’s implement useless Python objects(PyCon APAC 2023)
› ΈࠐΈؔ pow ͷΒΕ͟ΔਐԽ (PyCon JP 2021)
› ΠϯϝϞϦʔετϦʔϜ׆༻ज़ (PyCon JP 2020)
› ܅ cmath Λ͍ͬͯΔ͔ (PyCon mini Shizuoka 2020)
› Python ͱָ͠Ήॳ (PyCon mini Hiroshima 2019)
› SymPy ʹΑΔࣜॲཧ (PyCon JP 2018)
Ұཡ https://xaro.hatenablog.jp/ Λࢀর͍ͯͩ͘͠͞
5 / 34
Slide 6
Slide 6 text
Today’s Theme
໌͔Β graphlibɺΈΜͳͰ͓͏
6 / 34
Slide 7
Slide 7 text
Executive Summary
͍͠ਓ͚ͷཁ
› τϙϩδΧϧιʔτͱɺ༗ඇ८ճάϥϑͷू߹ͷઢܕॱংͰ͋Δ
› graphlib ɺτϙϩδΧϧιʔτ͕࣮͞Εͨඪ४ϥΠϒϥϦͰ͋Δ
› graphlib ɺ؆қతͳλεΫϥϯφʔͱͯ͑͠Δ
7 / 34
Slide 8
Slide 8 text
άϥϑͬͯɺԿʁ
ఆٛ (ແάϥϑ)
༗ݶू߹ V ͓Αͼ V ˆ V ͷඇॱংର͔ΒͳΔू߹ͷ෦ू߹ E ͷ G = (V; E)
ΛແάϥϑͱݺͿɻ
ఆٛ (༗άϥϑ)
༗ݶू߹ V ͓Αͼ V ˆ V ͷॱংର͔ΒͳΔू߹ͷ෦ू߹ E ͷ G = (V; E)
Λ༗άϥϑͱݺͿɻ
› V Λू߹ɺE Λลू߹ͱݺͿɻ
› V ͷཁૉΛ G ͷɺE ͷཁૉΛ G ͷลͱݺͿɻ
8 / 34
Slide 9
Slide 9 text
ແάϥϑͷྫ
9 / 34
Slide 10
Slide 10 text
༗άϥϑͷྫ
10 / 34
Slide 11
Slide 11 text
ͭ·ΓʜͲ͏͍͏͜ͱͩͬͯΑʁ
ʜͰɺͦͷάϥϑࣾͰಇ͘͏͑Ͱ
ԿͷϝϦοτ͕͋Δͱ͓ߟ͑Ͱ͔͢ʁ
11 / 34
Slide 12
Slide 12 text
དྷ͍Αϕωοτʂ ఆٛͳΜ͔ࣺ͔͔ͯͯͬͯདྷ͍ʂ
͍͠ਓ͚ͷάϥϑ
› άϥϑͱɺ
ʮͷʯͱͦͷʮؔʯΛֶతʹϞσϧԽͨ͠ͷͰ͋Δ
› ʮͷʯਓؒɺӺɺλεΫɺαʔόͳͲ
› ʮؔʯਓؒؔɺઢ࿏ɺґଘؔɺωοτϫʔΫͳͲ
12 / 34
Slide 13
Slide 13 text
τϙϩδΧϧιʔτͬͯɺԿʁ
ఆٛ (ೋ߲ؔ)
ू߹ A ͷੵ A ˆ A ͷ෦ू߹ R Λೋ߲ؔͱݺͿɻ
·ͨɺ(a; b) 2 R Λ aRb ͱද͢ɻ
ఆٛ (ॱংؔ)
ҎԼͷੑ࣭Λຬͨؔ͢ R ΛॱংؔͱݺͿɻ
ࣹ 8a 2 A ʹରͯ͠ɺaRa Ͱ͋Δ
ରশ a; b 2 A ʹରͯ͠ɺaRb ͔ͭ bRa ͳΒ a = b Ͱ͋Δ
ਪҠ a; b; c 2 A ʹରͯ͠ɺaRb ͔ͭ bRc ͳΒ aRc Ͱ͋Δ
ॱংؔɺେখؔൺֱͷ֓೦Λநͨ͠ͷɻ
13 / 34
Slide 14
Slide 14 text
τϙϩδΧϧιʔτͬͯɺԿʁ
ఆٛ (ઢܕॱং)
ू߹ A ͷॱংؔ R ʹ͓͍ͯɺ8a; b 2 A ʹରͯ͠ aRb ·ͨ bRa ͕Γཱ
ͭͳΒɺR ઢܕॱংͰ͋ΔͱݺͿɻ
ఆٛ (τϙϩδΧϧιʔτ)
༗άϥϑ G = (V; E) ͷτϙϩδΧϧιʔτͱɺV ͷઢܕॱং (V; ») Ͱɺ
(u; v) 2 E ͳΒ u » v Λຬͨ͢ͷͰ͋Δɻ
14 / 34
Slide 15
Slide 15 text
τϙϩδΧϧιʔτͬͯɺԿʁ
ఆٛ (าಓ)
άϥϑ G = (V; E) ͷ u1
͔Β uk
ͷ͞ k ͷาಓ W ͱɺͷྻ
(u1
; u2
; : : : ; uk
) Ͱɺ(ui
; uj
) 2 E(1 » i < j » k) Λຬͨ͢ͷͰ͋Δɻಛ
ʹɺu1
= uk
ͷ߹ɺW ด͍ͯ͡ΔͱݺͿɻ
ఆٛ (ಓ)
άϥϑ G = (V; E) ͷาಓ W ʹ͓͍ͯɺ͓Αͼล͕ͯ͢ҟͳΔͷಓͱݺ
Ϳɻಛʹɺดͨ͡ಓΛด࿏ͱݴ͏ɻ
ఆٛ (༗ඇ८ճάϥϑ)
༗άϥϑ G = (V; E) ʹ͓͍ͯɺด࿏Λؚ·ͳ͍ͷΛ༗ඇ८ճάϥϑͱݺͿɻ
15 / 34
Slide 16
Slide 16 text
τϙϩδΧϧιʔτͬͯɺԿʁ
໋ (τϙϩδΧϧιʔτՄೳ)
༗άϥϑ G ͕τϙϩδΧϧιʔτՄೳͰ͋ΔͨΊͷඞཁे݅ɺ
༗άϥϑ G ͕༗ඇ८ճάϥϑͰ͋Δ͜ͱͰ͋Δɻ
16 / 34
Slide 17
Slide 17 text
ָ͍͠ূ໌ίʔφʔ
τϙϩδΧϧιʔτՄೳ ) ༗ඇ८ճάϥϑ.
τϙϩδΧϧιʔτՄೳ͕ͩάϥϑ G ʹด࿏͕ଘࡏ͢ΔͱԾఆ͢Δɻ
ͦͷด࿏Λ vi
; vj
; vk
(i < j < k) ͱͯ͠ɺ͔ͭ vi
» vj
» vk
ͱ͢Δɻ
vi
; vj
; vk
(i < j < k) ด࿏ͳͷͰɺvk
͔Β vi
ͷล͕ଘࡏ͢Δɻ
ͭ·Γɺvk
» vi
ͱͳΔ͕ɺͦΕԾఆ vi
» vj
» vk
ʹໃ६͢Δɻ
17 / 34
Slide 18
Slide 18 text
·ͩ·ͩଓָ͍ͧ͘͠ূ໌ίʔφʔ
ิ
༗ඇ८ճάϥϑʹೖ࣍ 0 ͷ͕ඞͣଘࡏ͢Δɻ
ূ໌.
༗ඇ८ճάϥϑͷ࠷ͷಓ v1
; : : : ; vk
Λ 1 ͭબͿɻv1
ʹೖΔล͕ଘࡏ͢ΔͳΒ
ɺu; v1
; : : : ; vk
͔ͭ (u; v1
) 2 E ͷΑ͏ͳ u ͕ଘࡏ͢Δ͜ͱʹͳΔ͕ɺ
v1
; : : : ; vk
͕࠷ͷಓͰ͋Δͱ͍͏Ծఆʹ͢ΔɻΑͬͯɺv1
ͷೖ࣍ 0 ͱͳ
Γɺ༗ඇ८ճάϥϑʹೖ࣍ 0 ͷ͕ඞͣଘࡏ͢Δɻ
18 / 34
Slide 19
Slide 19 text
͑ʁ ·ͩ͋ΔΜͰ͔͢ʁ ָ͍͠ূ໌ίʔφʔ
ิ
༗ඇ८ճάϥϑ G ͷ෦άϥϑ H ·ͨ༗ඇ८ճάϥϑͰ͋Δɻ
ূ໌.
෦άϥϑ H ͕༗ඇ८ճάϥϑͰͳ͍ͱԾఆ͢ΔɻH ͷด࿏Λ C ͱͨ͠߹ɺ
H G ͷ෦άϥϑͳͷͰɺด࿏ C G ʹଘࡏ͢Δ͜ͱʹͳΔɻ͔͠͠ɺͦΕ
G ͕༗ඇ८ճάϥϑͰ͋Δͱ͍͏Ծఆʹ͢Δɻ
19 / 34
Slide 20
Slide 20 text
ָ͍͠Αͳʂ ূ໌ίʔφʔ
༗ඇ८ճάϥϑ ) τϙϩδΧϧιʔτՄೳ.
༗ඇ८ճάϥϑ G ʹඞͣೖ࣍ 0 ͷ͕ඞͣଘࡏ͢ΔͷͰɺͦͷ͓Αͼͦ
͔͜Βग़ΔลΛऔΓআ͘ɻऔΓআ͍ͨάϥϑ·ͨ༗ඇ८ճάϥϑͳͷͰɺ͕ͳ
͘ͳΔ·ͰͦΕΛ܁Γฦ͢ɻऔΓআ͍ͨॱ൪ʹΛฒͨͷ͕τϙϩδΧϧιʔ
τͰ͋Δɻ
ॏཁɿূ໌͕ΞϧΰϦζϜʹͳ͍ͬͯΔ͜ͱʹҙɻ
20 / 34
Slide 21
Slide 21 text
ࠓ PyCon JP 2025 ͷ 1 Ͱ͢
ࢄֶͷߨٛͰͳ͘ɺ
PyCon JP Ͱ͢Α
21 / 34
Slide 22
Slide 22 text
དྷ͍Αϕωοτʂ ূ໌ͳΜ͔ࣺͯͯʢ͈́
͍͠ਓ͚ͷτϙϩδΧϧιʔτ
› ॱংؔɺେখؔแؚؔΛநԽͨ͠ͷ
› ઢܕॱংɺͲΕͰൺֱͰ͖Δͭɺ͙Β͍ͷཧղͰ
› τϙϩδΧϧιʔτɺ༗άϥϑͷΛ͍͍ײ͡ʹॱং͚ͨ͠ͷ
› τϙϩδΧϧιʔτͱ༗ඇ८ճάϥϑදཪҰମ
› τϙϩδΧϧιʔτʹؔ͢Δূ໌͕ΞϧΰϦζϜʹͳ͍ͬͯΔ
22 / 34
Slide 23
Slide 23 text
τϙϩδΧϧιʔτͷ͍ಓͱ
Q: ΛλεΫɺลΛλεΫͷґଘؔͱ͢Δɻ
Ͳͷॱ൪ͰλεΫΛ࣮ߦ͢ΕΑ͍͔ʁ
23 / 34
Slide 24
Slide 24 text
τϙϩδΧϧιʔτͷ͍ಓͱ
Ans: άϥϑΛτϙϩδΧϧιʔτͯ͠ɺͦͷॱ൪ʹΕΑ͍ɻ
A ˠ B ˠ D ˠ E ˠ C ˠ G ˠ F ˠ I ˠ M ˠ H ˠ J ˠ K ˠ L ˠ N
24 / 34
Slide 25
Slide 25 text
Ͳ͏ͬͨͷʁ
͢Ͱʹ graphlibɺͬͨΑ
25 / 34
Slide 26
Slide 26 text
ਅଧొ
graphlibɿάϥϑߏΛૢ࡞͢Δػೳ
› ݱࡏ TopologicalSorter ͷΈɺ୯७ͳඪ४ϥΠϒϥϦ
› Python 3.9 ͰՃ
› https://github.com/python/cpython/issues/61207 Ͱٞ
͞Ε͍ͯͨɻ
26 / 34
Slide 27
Slide 27 text
graphlib ͷ͍ํ
τϙϩδΧϧιʔτͷ࣮ߦ
from graphlib import TopologicalSorter
graph = {
"D": {"B", "C"},
"C": {"A"},
"B": {"A"},
}
order = TopologicalSorter(graph).static_order()
print(f"Topological order: {" ˠ ".join(order)}")
# A ˠ C ˠ B ˠ D
D B, D C ͷΑ͏ͳΠϝʔδͰఆٛ͢Δɻ
27 / 34
Slide 28
Slide 28 text
graphlib ͷ͍ํ
༗ඇ८ճάϥϑͰͳ͍έʔε
from graphlib import TopologicalSorter
graph = {
"A": {"C"},
"B": {"A"},
"C": {"B"}
}
order = TopologicalSorter(graph).static_order()
print(f"Topological order: {" ˠ ".join(order)}")
# graphlib.CycleError ͕ൃੜ͢Δ
28 / 34
Slide 29
Slide 29 text
ΑΓߴͳ͍ํ
graphlibɿґଘؔͷ͋ΔλεΫΛ࣮ߦ͢ΔλεΫϥϯφʔ
› ฒྻ࣮ߦ͍͢͠ΞϧΰϦζϜ
› υΩϡϝϯτʹͦΕͱͳ͘৮ΕΒΕ͍ͯΔ͕ɺෆे...ɻ
› ੜ AI ͷྗΛआΓͯͦΕͱͳ͘࡞ͬͯΈΑ͏ʂ
29 / 34
Slide 30
Slide 30 text
λεΫͷఆٛ
dataclass ʹΑΔλεΫఆٛ
@dataclass(frozen=True, slots=True)
class Task:
name: str
action: Action
deps: set[str] = field(default_factory=set)
ϋογϡՄೳͰ͋Δඞཁ͕͋ΔͷͰɺfrozen=True Λ͏ɻ
30 / 34
Slide 31
Slide 31 text
λεΫϥϯφʔ
άϥϑͷఆٛ
def run(tasks, max_workers=None):
by_name = {t.name: t for t in tasks}
ts = TopologicalSorter({t.name: set(t.deps) for t in
ts.prepare()
results: dict[str, object] = {}
...
31 / 34
Slide 32
Slide 32 text
λεΫϥϯφʔ
ฒߦॲཧɿεϨουʹೖ
def run(tasks, max_workers=None):
...
with ThreadPoolExecutor(max_workers) as pool:
while ts.is_active():
ready = ts.get_ready()
futs = {}
for name in ready:
print(f"run {name}")
futs[pool.submit(by_name[name].action)]
...
32 / 34
Slide 33
Slide 33 text
λεΫϥϯφʔ
ฒߦॲཧɿྃͪ
def run(tasks, max_workers=None):
...
with ThreadPoolExecutor(max_workers) as pool:
while ts.is_active():
...
for fut in as_completed(futs):
name = futs[fut]
results[name] = fut.result()
print(f"done {name}")
ts.done(name)
return results
33 / 34
Slide 34
Slide 34 text
·ͱΊ
·ͱΊ
› τϙϩδΧϧιʔτͱɺ༗ඇ८ճάϥϑͷू߹ͷઢܕॱংͰ͋Δɻ
› ༗άϥϑ G ͕τϙϩδΧϧιʔτՄೳͰ͋ΔͨΊͷඞཁे݅ɺ
༗άϥϑ G ͕༗ඇ८ճάϥϑͰ͋Δ͜ͱͰ͋Δɻ
› graphlib ɺτϙϩδΧϧιʔτ͕࣮͞Εͨඪ४ϥΠϒϥϦͰ͋Δɻ
› graphlib ɺ؆қతͳλεΫϥϯφʔΛ࣮͢Δࡍʹɺ࣮ߦॱংΛܾఆ͢Δ
Έͱͯ͑͠Δɻ
34 / 34